Senate Transportation Committee
No chapters available yet
Legend
Transcript markers
- text Sentence being played
- text Search result
- text Intervention cited by the link you opened
Click a sentence to play the video from that moment. Select text to quote it with the speaker, timestamp and link.
More options
Senate Transportation Committee California State Senate
Follow its sittingsSpeaker 3
Thank you. You. Just making sure this is okay, great. Chair Cortese, members of the committee, my name is Ariel Wolf. I serve as General Counsel to the Autonomous Vehicle Industry Association, which represents the world's leading autonomous vehicle companies deploying AVs across the country.
Thank you very much for having me here today to discuss the state of the AV industry and its role in California's transportation future. I have a few slides to accompany my remarks. I'm, of course, eager to answer any questions. You can stay as long as you need for that. First, shown here is Avia's membership.
We cover the gamut of companies that are building the autonomous vehicle future. Light duty and heavy duty vehicles, robotaxis, delivery vehicles, auto manufacturers, hardware and software suppliers, and others. I say all that because it's important to understand the expanding ecosystem that is flourishing across The United States and here in California today due to the ingenuity of our nation's innovators and the promise of AV technology to solve real problems. Before we get deeper though, I wanna make one key point to avoid any confusion. People talk about the six levels of autonomy, and I will show that in a moment, but there are really just two technologies to focus on for this discussion.
One, course, is autonomous vehicles and two, vehicles with driver assist technology. Autonomous vehicles drive the car without human driver intervention. In AVs, humans are the passengers, not the drivers. Driver assist technology requires a human driver to remain fully engaged, monitor the roadway, and be ready to take control of the vehicle at any moment. This tech is widely available in consumer vehicles today, like lane keep assist, automatic emergency braking, and adaptive cruise control.
You must be attentive at the wheel and keep your hands and eyes forward. This distinction matters both for safety and for public understanding. If a human has to be attentive at the wheel at all times, it is not an AV, regardless of how it's described or marketed. And you may have seen these levels of automation before, in the slide here. Given the time constraints, I won't go through this now, but I'm happy to return to this at a later time for any questions.
Here, in the slide is a generalized description of an AV. It shows the many sensors and technologies that are used by the automated driving system, or the ADS, to perform the driving task. One quick note here is that these sensors see beyond what the human eye can perceive in a many and in many directions all at once. The ability to synthesize all of this information simultaneously is what gives an AV a full picture of its surroundings. This is an image from a LiDAR unit, on the left, a light infrared detection light detection and ranging, I should say.
And then the one next to it is from a vehicle's object classification function. And the next few slides focus on the crisis we have on our roads today and will frame the rest of my remarks. Two weeks ago, this committee held a joint informational hearing that examined recent crash data. The background memo for that hearing made clear that, quote, road injury is consistently the leading cause of death for children and young people ages five to 24 in California. There were seventeen thousand seven hundred and forty five fatal and serious injury crashes in California in 2024, and California's rate of fatal and serious injury crashes is still far higher than it was ten years ago.
The committee's memo emphasized the central role of human behavior in these crashes, DUIs, distraction, fatigue, and excessive speed. And I would direct you back to that memo that lays out the grim statistics of human caused fatalities. AV deployment is not an abstract policy debate. It is a response to a present and urgent public safety failure caused by human behavior. Some data already suggest that AVs have reduced serious injuries and crashes by ninety two percent.
We can either continue to rely on attempts to manage human behavior, or we can embrace technologies designed to remove the risks that our current system has struggled to control. AVs operate today on public roads across the country as shown here, including here in California with a strong and growing safety record. They are transporting passengers and delivering goods commercially in places like California, Arizona, Nevada, Texas, and Michigan. The workforce opportunities created by the AV industry are diverse and expanding. These include service technicians, remote support and delivery personnel, and data collection specialists.
Many of these roles do not require a four year college degree, but do offer pathways to stable, high quality employment. The AV industry also is an economic engine for California. In fact, one of our members, Zoox, has opened the first ever serial production facility for purpose built, all electric robotaxis in The US and will create hundreds of jobs in the San Francisco Bay Area. California has has been the birthplace of this industry and remains a global leader. Right now, the California DMV is finalizing updates to its AV regulations.
The DMV's AV regulations already are the most stringent in the country.
Speaker 3
AVs represent transformative advancement in transportation, one that is already saving lives and reinforcing California's position as a global leader in innovation. The AV industry is eager to work with the committee and all policymakers to achieve this vision. Thank you very much and I look forward to your questions.
Speaker 1
Thank you very much. Apologize for the pause there. We've competing committees calling over here. We'll move on now to Doctor. Missy Cummings, who's with us virtually.
Doctor.
Speaker 4
Thank you. Thank you for the opportunity to speak to you today. I do want to recognize that there are a lot of positive benefits to autonomous vehicles. We may argue that driving assist systems, I do think they count as partial autonomy, which we'll talk a little bit about later. And, indeed, I wanna start there.
I have three very significant concerns, and I will first address the driving assist systems. They, as well as self driving cars, rely on computer vision. This is very important because computer vision is a very unreliable technology, and it's especially a problem because unreliable technology in any autonomous vehicle, ADAS or self driving car, if it's unreliable and you rely on a human, but humans are, have a tendency towards complacency in vehicles, then it's especially problematic when the autonomy fails and then the human is not paying attention. You're gonna hear more about that from another witness today, but these problems mean that no ADAS vehicle should ever be allowed to operate in any operational design domain for what which it was not explicitly designed for. Moving on to my second point.
You've heard a lot here today about self driving cars. There isn't such thing as a self driving car anywhere in the world. All of them. All of them. All of them in California require significant human babysitting in the form of remote operations.
And California, your self driving cars are being controlled from people in The Philippines, people who do not have US driver's license licenses. This is very unsafe because the latency in the signals between California and The Philippines has and will continue to provide unsafe situations that puts the public at risk. These people need to be brought back to The United States, and they need to have US driver's licenses. My last point, it actually involves both ADAS and self driving cars. And this revolves around, yet again, computer vision, a very unreliable technology, but this problem is the problem of hallucinations.
These cars, again, cars with ADAS and self driving cars, hallucinate. They see things that aren't there. This will often result in an emergency hard breaking maneuver, which can and has led to accidents, can and will continue to kill people. If the state of California does not do something about addressing this phantom breaking problem and you let self driving trucks roll out on your highways, I assure you people will die. The people of California have been guinea pigs for the self driving car community, And there are very serious problems, which we have data on, and we need more data to address.
But we can't ignore these problems, and the public in California shouldn't be made to pay that price. I yield my time.
Speaker 1
Alright. Thank you very much. So we'll move, to the next individual testifying, Robert O'Dowd. Welcome.
Speaker 5
Chairman and members of this committee, thank you for giving me the opportunity to speak today. My name is Robert O'Dad, and I'm a member of the DOM Project, a public safety advocacy organization dedicated to making computers safe for humanity. And I am not just simply an unaffected advocate. I'm also the father of a two year old. And no parent should ever have to hear that their child was struck by autonomous driving vehicle with software that was known to be defective and then have to fight through years litigation against corporate obstruction in order to reveal crash data that was hiding on their servers for years.
I know that this is possible because a victim in Willow is here with us today. Allow me to explain. Current law permits any company to deploy AI at scale on the roads with little to no oversight. Tesla Motors does this under the product names of full self driving and autopilot. This trail blazes a dangerous path for other actors to imitate.
The incentives are clear. Pushing autonomous driving intelligence onto our roads by disrupting labor and transportation and then selling it to safety conscious Californians by repeating that it is safer than a human driver right now today. However, the evidence has borne out that it is not, in fact, safer. So they have to exploit gaps in the law in order to avoid regulatory oversight and investigations into how their system is works and designs. In the event of an accident, the manufacturer is allowed to redact information about the software's implications in the accident and then place all the remaining blame on the human operators.
Any safety statistics statistics regarding the operation of these vehicles comes through their own corporate marketing functions, cherry picked and manipulated. Help us fix this problem by addressing these gaps in the law that allow self driving software to operate anywhere with no independent safety validation. Victims harmed by these accidents by the corporate Goliath in an uphill battle as they profit from selling this technology with no obligation to the truth. By far, the worst offender of all of this manipulation is Tesla. To date, Tesla has reported fifty nine fifty nine fatalities to this to NHTSA involving full self driving and autopilot.
Over half of those fatalities were right here in California. We have provided video evidence that Tesla's running using full self driving technology. Ignore school zones, stop signs, and closed roads. We have verified reports of Tesla's using these features colliding into emergency vehicles and first responders. However, under current law, fixing these mistakes are voluntary and at Tesla's own convenience.
These are not edge cases. The scope and scale of these appointments are immature or premature, and this constitutes a public safety hazard. Tesla promotes self full self driving by frequently promoting that it will be ready, fully autonomous by the end of this year, a line that they have repeated every year for the last decade. But a nearly complete project should not still continue to have flagrant design omissions as well as safety critical defects. Litigation has revealed that Tesla knows about these capability gaps yet fights to conceal auditable safety statistics.
However, at this moment, I wanna focus on one inexcusable failure. Consider the case of Tillman Mitchell. In November 2023, the DOM project published its scathing report to the New York Times that the AIs and Tesla could not recognize the school buses with their stop signs extended. We informed Tesla, we informed the regulators, and no changes were made. Five months later, Mitchell was hit as he was exiting his school bus by a Tesla running autopilot, sending a child to the hospital with life changing injuries.
Now this may come as a surprise, but these defects are still present in diversions on the roads today. And under current law, Tesla has no obligations to address them, opting to blame the humans every single time these defects manifest. It is unconscionable to allow millions of users to unwillingly enable a defective AI with crucial fixtures without crucial fixes. I ask you, banish this from our roads, never to see the light of day again. It has only been through the victims such as Dylan here today and during years of litigation that we can establish this pattern of malfeasance.
You have an opportunity to ensure that no more families need to fight to uncover information that is in the public's interest. We have two asks of this committee. Force any program controlling a vehicle to comply with level four autonomous vehicle disclosure standards. No more ability to hide no more ability to redact to hide failures. No more hiding behind the legal fiction that is simply a level two driver assistant in order to place blame on victims.
Second, mandate the prompt resolution of known safety defects in any software controlling steering and braking. Where these vehicles cannot be fixed, the software must be disabled. We cannot allow known defective systems to remain active on our roads any longer. Members of this committee, Tesla operates under the arrogant assumption that this committee will change nothing and permit them to continue obstructing justice for their victims. The delay in justice is itself an injustice.
Thank you, and I look forward to answering your questions.
Speaker 1
Well, thank you very much. Appreciate everyone working hard to abide by the time limitations. I'm sure everyone else does as well. Mr. Angelo, welcome.
Appreciate you being here. And you can start anytime you want. Mr. Chairman,
Speaker 6
the mic's not working. Mister chairman, mister chairman and members of the committee, thank you for giving me the chance to speak today. My name is Dylan Angulo, and I flew here all the way from Florida for one reason, because I still believe my pain in Nibel's life can mean something if it helps protect other families from going through what we went through. I came here to honor Nibel Benavides Leon, the woman I loved whose life was taken in an instant. And I came here to stand up for every victim and every family who should not have to bury someone or live the rest of their life broken because dangerous technology was allowed on public roads without enough rules, transparency, and accountability.
On 04/25/2019, Nibel and I were standing on the shoulder of a road in Key Largo, Florida. A Tesla Model s operating with autopilot engaged came through a t intersection at about 62 miles per hour. The evidence showed the car recognized a stop sign, recognized us pedestrians, recognized the end of drivable space, and recognized our parked vehicle off the road, yet it still went straight through without stopping or warning the driver in time. Nybelle was killed. I survived, but I did not walk away whole.
I suffered a traumatic brain injury, a shattered pelvis, a fractured sacrum, and other severe injuries that changed my life forever. I live with permanent pain and I walk away. I walk I walk like a broken old man. So when this technology is described as innovation, driver convenience, and the future of transportation, it is important to understand what that sounds like to a person like me. This is not a theory.
This is not a branding issue. This is a death. This is a permanent injury. This is a real human cost. And that is why I came all the way here asking for action before more families are destroyed.
Tesla has has had these vehicles on public roads for about a decade. The state may classify them as level two systems, but they are being marketed in ways that create the impression of full autonomy. Meanwhile, the public and Tesla's own customers have been unwilling participants in Tesla's real world self driving testing. Our public roads have become Tesla's test track. Tesla rushed this unfinished technology onto public roads while trying to lead the race to develop autonomous vehicles without the safeguard the public deserves.
And in court, we proved why that is so dangerous. We proved these vehicles were not as safe as Tesla claimed. We proved Tesla was not transparent with critical data, including failing to turn over the augmented video and crash data from the autopilot computer. We proved Tesla claims that these vehicles were safer than human drivers was not truthful. We proved Tesla ignored safety regulators and failed to meaningfully geofence or limit this technology.
That is why this is bigger than my case. Sadly, Nibel's story is not the only one. California knows this issue is not theoretical, and that is exactly why this state should be paying close attention and leading the way on regulation and accountability. To a pedestrian standing on the roadside, there is no meaningful difference between a so called driver assistance system and a more advanced autonomous system. If the vehicle fails to brake when a human life is in front of it, the labels don't matter when the outcome is the same.
And if this gap remains open, other families will go through what we went through. So I respectfully ask this committee and the state of California to help make sure Nybele's life leads to change by regulating this technology before more lives are destroyed. Please require real transparency when these systems crash. Please require preservation and disclosure of vehicle data after serious collisions. Please require safety limits, including restricting these systems from roads and conditions where they have not actually been proven safe, and please hold these comp these companies accountable when they put the public at risk.
This is not anti technology. This is pro human life. That is what responsible leadership looks like. Nivell never got a second chance. She never got to go home.
She never got to grow older. She never got to live the life she deserved. I did get a second chance, but I have to live every day with the pain, trauma, and damage this technology left behind. So I am asking respectfully but urgently, please do not let more families pay this price before meaningful action is taken. Please hold these companies accountable, and please make sure Nybele's life and the suffering of other victims lead to change.
Thank you.
Speaker 1
All right. Thank you, Mr. Angola. I apologize for mispronouncing your name earlier. It's the least I could do, having traveled here 3,000 plus miles.
We appreciate you being here, as I said earlier. We will bring it back to the committee at this time. And I do want, in case anyone came in after the introductions, to point out or remind that Doctor. Missy Cummings is available for Q and A as well on the monitor. She's here virtually.
So start with Vice Chair Strickland.
Speaker 7
Thank you, Mr. Chairman. Sorry about your loss. I know how hard that is. My best friend, when I was in fifth grade, got hit by a car on a bike, lost my best friend in fifth grade, and my teammate, my basketball teammate in tenth grade got in a car wreck.
Car wrecks are just there's no words I could do moving forward, but I do have some questions for mister Wolf. Look. When you drive a car, it's an incredible responsibility, and accidents are happening all the time. Is there data that talks about people who drive their cars themselves versus the AV cars? And do you have data that shows maybe one having more accidents than the other?
Is there data out there that would enlighten us as a committee?
Speaker 3
Yeah. Vice Chair Strickland, thanks for the question. And I'll associate myself with your comments as well. And I have four daughters, and it's terrible to hear tragedies. Absolutely.
In terms of the data, think it is very important to point that out, and the stories, of course, that you noted are compelling. The data very clearly shows, as I noted in some of the slides, from the committee's own work a couple weeks ago, that human impairment, distraction, intoxication, drug driving, fatigue, those are the primary drivers of collisions and fatalities and injuries. And I think, you know, the number that's used a lot, because it's true, is that we average around forty thousand fatalities, traffic fatalities, a year, overwhelmingly caused by human impairment. What it doesn't talk about, as was noted here, the injuries on top of Hundreds of thousands, millions of injuries, life altering injuries, of course. And so, those are the numbers.
And the statistic, ninety four percent has been used for some time for human impairment. It may be a little bit less than that, but regardless, the point is straightforward. Human behavior is the primary cause of the carnage on the roads. And so, the question then becomes, can technology play a role in solving that? Some initial data.
We're going be coming out in just a little bit of time with a new number of miles that autonomous vehicles have been on public roads in commercial service. It's in the hundreds of millions of miles over the past number of years, and the data is showing that it's having a dramatic positive impact, up to even ninety two percent reduction in fatalities, eighty percent reduction in collisions in within the areas that they are present, of course. And so I think that starts to show something, that we need to do there. And I just want to make one final point there is where I started, in my opening remarks. That is talking about level four fully autonomous vehicles, not vehicles that expect to have an attentive human, licensed human driver at the wheel with other technology around it.
That is a different type of technology. It's not an autonomous vehicle, and there's some discussion of that. But I want to make that distinction very clearly that on the autonomous vehicle side, fully autonomous vehicle side, it's having a dramatically positive impact on safety.
Speaker 7
Well, a follow-up question. The testimony from is it doctor Cummings or miss it's doctor. Doctor Cummings. She said that your autonomous vehicles are not done by computers but by people in The Philippines. What's your response to that?
Because you said you're trying to get away from human error, but if humans are doing it in The Philippines, you would still have the human error, wouldn't you?
Speaker 3
And I would, of course, direct questions to the specific companies that are talking about that particular note about Philippines. But I think, as a general matter, there's two kinds of remote operations. One is remote assistance and one is remote driving. Maybe take the latter. Remote driving would actually be driving the vehicle remotely, of with a teleoperation is another term for it.
That's not what we describe with a level four autonomous vehicle. Vehicle.
Speaker 7
So that would not be the case for level four. No one from The Philippines is controlling?
Speaker 3
My understanding, sir, that that's not the case. That remote, remote assistance would be the vehicle remains the the vehicle is performing the driving task. All aspects of the driving task. It may encounter a situation that then gets referred to a remote assistant to make a decision, a binary decision, yes or no. Sort of a sliding scale of how that remote assistance gets involved.
But at no point in that remote assistance is that performing the driving task. That would be a different, a different set, you know, a lower level of autonomy. So in in fully autonomous vehicles, it would be remote assistance. There are different ways of doing it. I I know for some companies, it's all being done here in The United States.
There's other companies can answer for themselves. But that's the the essence of the distinction.
Speaker 7
So my understanding is Tesla performs the same functions as the autonomous system, steering, braking, navigation. Why should it be exempt from the same testing reporting standards that are required for companies like Waymo? What would be your response to that? Well, the outset, would
Speaker 3
just say Tesla is not a member of our association. I don't speak for Tesla, so direct any questions to the company. What I can say is that in the state of California, it's required to obtain a permit to put level three or above fully autonomous vehicles on the roads, and it's my understanding that Tesla does not do that.
Speaker 7
Okay. Back to Mr. O'Dowd. Mr. O'Dowd, you mentioned, and I didn't catch it, Tesla had a certain amount of fatalities.
What was the number? I was trying to write it down.
Speaker 7
Fifty nine. And have you measured that versus normal accidents with, you know, just say a a Toyota or how many accidents are happened from Tesla versus some of the other cars that are on the road?
Speaker 5
In absolute scale or in relative to Relative. You know, obviously, I wanna compare apples to apples. It is difficult to say because these are accidents that are attributable to FSD or to autopilot, but there isn't a comparable autonomous level two type of system that operates in the same domain that would fall under the same classification. You would have to take a look at the total number of fatalities attributed to either one of the manufacturers and then drill down into how many of them have reported their level two systems because that is how Tesla categorizes and compare them. When we have done the statistic, Tesla makes up the overwhelming bulk of the level two fatalities.
I would I could get back to you with a precise number, but it is larger than any other manufacturer.
Speaker 7
interested. Again, we have to compare apples to apples. If if a level two Tesla has so many fatalities and you have the same amount of cars on the other side, what would that comparison be? Because we're not starting from a position of zero.
Speaker 5
Yes. In absolute count, it is the majority of level two fatalities are attributable to Tesla.
Speaker 7
So last one for mister Dowd. You mentioned some of these failed to stop. From your experience or from your expertise, is it a a learning curve issue, or is it a fundamental failure with the software from from your perspective?
Speaker 5
My perspective on this particular matter is that Tesla has been working using this methodology of collecting data through successful and also failed incidents where it is successfully sought at stop signs and to promote those types of learnings within their AI. And then to take every time that it has failed to stop for a stop sign and then use those to penalize the AI. Over years and after millions and billions of miles, it has tried training an AI that can try to sort out the situations where it should stop and where it should not stop. This has yielded some results, and that is the product of the vehicle down on the roads today. However, right now, I can give you a 100% guarantee, and you can even drive in the car with the current updated version, that it will still make these mistakes easily and replicably.
There are roads within my own backyard where I can go, and it will 100% of the time demonstrate failures to recognize basic signage on the roads and basic road rules. I would say that this is a failure of capability and approach. I do not believe that more data or a more powerful hardware stack will solve these at a fundamental level. There are far more details that could be gone into, and I'd be happy to go into the
Speaker 7
And on that statement, are you talking about level four and level two or just level two?
Speaker 5
My particular stance in this matter is that if the car is capable of setting navigation and then you could release your hands from the wheel and go from location to location, it would fall under what we would imagine is the level four designation. But Tesla does label their system as a level two system legally, and that feels like a bit of an issue because if it was required to be reported the level four expectations, that it would probably be very incapable as a level four, a fully autonomous vehicle.
Speaker 1
Okay. Thank you. Any other members of the committee? I have a couple of questions. Part of the difficulty in just following all the testimony and dialogue so far, and we'll get used to it, I guess, by the end of the committee hearing.
And I may be speaking for myself, but we've had testimony, if you will, from you, Mr. Wolf, that there are only two kinds of vehicles out there, autonomous and driver assist. But we have DMV itself using terms like FSD or FSD supervised. We've had, I thought, some good questions from the vice chair around whether or not there's driver assist available even with autonomous vehicles that you're distinguishing from driver assist. And it's my understanding that there's some level of driver assist available for all autonomous vehicles that are on the road today.
Is that incorrect?
Speaker 3
Thank you, mister chairman. I think it is very important to clarify this, and I I would agree. It it can be confusing, because, in part, perhaps because for a number of years, there was the SAE levels of automation, zero through through five, and that created a decision. That continues to be relevant for engineering. But the distinction between the two kinds of vehicles is really important.
Whether a human is participating in any way, even just an expectation of taking back control, that's the industry does not consider that an autonomous vehicle. And I think maybe the place to really draw the distinction is increasingly now because of federal government involvement, there are going to be vehicles, and there are some vehicles, that do not have any manual controls in the vehicle at all. And that, of course, is going be a fully autonomous vehicle. There's no ability to drive it other than the vehicle itself driving. So that's, as you continue to have some level four vehicles with manual controls because of a federal, motor vehicle safety standard issue, that's that presents some of the confusion.
But that dichotomy, I think, is very important to, to have in place that any company that is saying human licensed, even if they're really doing very little in the end, that more and more of the vehicle is performing a driving task, if the expectation is to be able to take back control at a second's notice and to be attentive to the road, it's assisting that driver. It's not the vehicle. It's the human's not the passenger. Humans are participating in that. And that that's important to retain as a distinction, and we feel as the autonomous vehicle industry.
Speaker 1
So am I am I mistaken in sort of in a simple it'll be an oversimplification of understanding the technology, but you have a coded electronic computerized system that's running a fully autonomous vehicle, as I understand you're defining it, where you have no, essentially no human able to take actual manual control at the time. But that coding in that programming was done by humans at some level, correct?
Speaker 1
And if I'm understanding the way this works, when that coded information has problems informing the autonomous vehicle as to how to navigate a situation, there's human assist on a remote level. Is that not true?
Speaker 3
Well, as an operational thank you for the question. There there there's an operational apparatus around the deployment of an autonomous fleet. That involves, of course, maintenance and and everything that goes into managing a fleet. In the case of autonomous vehicle operation, this is one of the interesting things about workforce development, there's an apparatus associated with the operations of that fleet. And that could involve, in a situation where a vehicle that is has been tested rigorously and over many, many millions of miles of simulation than on on road testing, and encounters a situation that requires, a human to point out something like that.
That that still does occur. Of course, that learning one interesting thing about the way that works is that learning in that instance gets pushed to all of the vehicles in the entire fleet for that autonomous vehicle. Unlike in the human driver situation, if we learn something in one case, it doesn't get imputed to all the other millions of drivers on the road as well. So it's actually so those situations help the vehicles learn and operations improve.
Speaker 1
That's good to know. And what I was really getting at is, your concerns, valid concerns, about human failure when it comes to controlling vehicles that are moving down the road at moderate to high speeds, making navigation decisions that I think kind of broad brushed a 92% safety increase with autonomous vehicles over human drivers. The question and that's because of elimination of things like human fatigue, intoxication, or substance abuse, and I forget what the other issue was. Wouldn't those same issues be of concern in a backroom operation with remote driver assist?
Speaker 3
Mr. Chairman, it's a great question. So, there's remote driving, which we said is not the case in autonomous vehicle. That those would present, certainly. I think there are workforce, issues to explore for remote operations.
And that's a discussion that's happening also Washington, same kind of line of questioning. I think it is important to look at those things. I know the companies have rigorous methods and workforce issues and operations that they put in place for those kinds of things. So I think it probably is a little bit different than an intoxicated person sitting behind the wheel of a car going 110 miles an hour on the road versus those situations. But to the extent that they require examination, I think the companies are forthcoming about that and certainly sharing information with NHTSA and other entities.
Speaker 1
And I think I'm asking you these questions because it seems to me that if there's genuine concern about eliminating human error, then it begs the question what are we doing to eliminate human error with autonomous vehicles as well. And what I guess I'd ask you as an attorney, and I don't know where your area of specialty is, I practice a little bit of tort law when I was practicing full time, shouldn't it be a concern, some level of concern for your members that there should be some standards for the behavior that's taking place with the human element that's backing up and informing these autonomous vehicles from just from a strict liability standpoint?
Speaker 3
It's certainly an important issue that's being looked at by all the companies. I think, as a general matter, tend to look at the liability and see it as an area that can well be accommodated by existing tort law and apportioning liability in that respect. I think in a situation where there's a company that owns and operates and manages the entire fleet and the vehicle's fully autonomous, the human has no role. In in some respects, depending on the facts of the case, it can be simpler than having then a human driver and trying to ascertain what happened. It would be the, you know, the automated driving system to the extent that there were facts to indicate it was at fault.
Of course, we see more, you know, overwhelmingly the opposite, that there's reason to believe that it's human drivers that are causing the issues and collisions. But to that extent, it can simpler.
Speaker 1
Yeah, and I get that. And again, just trying to bring it back to classic legislative regulation, which typically wants to make uniform standards to protect the public, but also to create fair play, to make it possible through mechanisms like insurance to make people whole if and when there are breakdowns, when there's tort awards, basically. And so really my question was, as you deal with all these, what we are currently talking about in terms of with an emerging technology, several different types of technology being used or utilized, but all of them, all of them at some level having a human oversight element to them. Should we be looking at trying to create uniform standards? You mentioned NHTSA, that's the federal side.
When I say we, I'm talking about California, the state of California. Obviously we have folks operating here right now. Especially in the field of robotaxis.
Speaker 3
Well, think it's an interesting point. I would say, just on the insurance point, I know Swiss Re has looked at, done some studies, has been able to corroborate a certain amount of data showing the diminution and the reduction of collisions and severe injuries and so forth as a result of AV deployment. I think topics like remote assistance, as long as I think it's contextualized as what it is and not conflated as can happen because of there's so many different things going on, conflated with road driving or even more significant road operations. Remote assistance is an important topic, and I know that it'd be important dialogue with DMV and the agencies and with lawmakers, of course, to help understand how we can provide assistance and information about how that works.
Speaker 1
And I appreciate the fact that, like so many things, the courts, based on current case law, can easily sort out liability and comparative negligence and who has to pay and who doesn't have to pay. But think, again, from a consumer protection standpoint, there's the question of how much can you do to head off those kind of claims in the first place. We heard some of that testimony today. Granted, I think the testimony today was relative to a technology that's not involved in your particular association, I understand that. So it's not a pointed question.
But it is a question about whether or not there should be more uniform standards than there is right now, whether or not that's a benefit to the industry at large.
Speaker 3
Mr. Chairman, just very quickly, I would just say the industry strongly supports uniform standards, certainly at you know, design, construction, and performance at the federal level, which helps ensure that we can drive a vehicle from Washington, D. C. To Sacramento, not get stopped in between, but, certainly other kinds of uniform standards, whether it's regulatory, legal regulatory, or engineering, or other kinds of policy, that is always helpful to not have that kind of patchwork of types of approaches.
Speaker 1
All right. Thank you very much for your candid responses. And I'd like to come back to you, I want to give you a chance to think about this one because I have questions for two other panelists right now. Anyway, the 8%, if you will, we talk about 92% improvement in safety. The 8% of non improvement, I'd love to come back to you in a moment before we excuse the panel and ask you to just think about who should be accountable for that 8%, given the pre autonomous vehicle standards that we have typically in the state that put points on a driver's license that can revoke people's driver's licenses.
On the 92% that are on the driver's side, it's pretty clear what the liability is, what the accountability is, what can be done from a state standpoint to get that person off the road, to impound the vehicle, etcetera. What should that look like with the 8%? And who is that you know, that becomes or should be accountable? Thank you. Let me go to Doctor.
Cummings just for a moment. Doctor. Cummings, if you're still there?
Speaker 1
Oh, great. I wanted to give you a chance to talk about comment that you made that there's no such thing as no human oversight or assistance with any of the technologies that we're talking about today. I hope I'm not paraphrasing you in a way that misrepresents what you said. But I think you know the comment I'm referring to. If you could just restate that and say more about that.
Speaker 4
Absolutely. So to be fair, Ariel is right that there are two kinds of general rote remote operations. One is remote driving, and that is like video game driving. The other is remote assist. But don't be fooled.
Remote assist is also remote control. So for example, today, self driving car companies, instead of driving the car with an accelerator and a brake pedal, they actually use keystrokes. They can tell a car, forward five feet at five miles an hour. Go forward a half a mile at 20 miles an hour. So I really appreciated your conversation about standards because we are in the wild, wild west of this.
It is not correct to say that remote assist is not control. It is control. And we need to understand that regardless of whether or not you're doing remote assist or remote driving, and the military learned this the hard way, you cannot do this from The Philippines safely. It incurs at least an additional two second latency delay. On average, it could be much worse.
And that two seconds is enough. It has caused an accident, at least one in California. There are others that are being looked at. So we know that the time delay from The Philippines, from the remote assistant to the cars operating in California can cause accidents. They have yet to cause deaths, but it's just a matter of time.
And
Speaker 1
with that said, do you have any thoughts on how to standardize that human element, how to begin that process? Policy wise, it's a common practice, for example, in the world we're in, to go look at best practices. I don't know if that's Waymo right now or somebody else, and start there and try to build upon good work that's already been done by the industry. Is is that the way to start here, or is academia the way to start? Or
Speaker 4
Well, you know, you're awesome. You're like my straight man. I just published an article a few days ago that said that the self driving car industry should look at best practices from the military and their remote drone operations. Indeed, there's a lot of lessons that they could and should be learning from that. I do not think that Waymo has the best practices.
Waymo is the company with the self driving, remote operations in The Philippines. This is unacceptable. As far as standards go, I did I was a senior staky adviser for NHTSA a few years ago, and I was trying to get my arms around the problem then. Federal government is slow to move. I absolutely think that this should be driven.
The lead should be taken by the federal government. They are being slow to act. California, it this is all happening in your backyard. I would wish that you would take the lead, make regulations that, legislation, that remote operation must bring those jobs back to California and also must make them have California driver's licenses.
Speaker 1
Mhmm. And and in some sense, just in fairness to Waymo, the general comment I was making or trying to make was they have a model and a system that essentially, again, from their own corporate standpoint requires the folks that are in The Philippines, I don't know that we throw that model out the window, but do we build on the practice and try to bring it to California standards, which is what I'm hearing you saying. But, again, I don't wanna put words in your mouth.
Speaker 4
company that I would point you to for effective remote operations would be Zooks before I would point you to Wayne.
Speaker 1
Alright. Fair enough. Let me turn to, mister Angulo. And you you very eloquently stated your case and also what happened in your actual case. At least you ran through some of the things that you were evidently able to prove at trial, as I understand your testimony.
What was the most challenging aspect of of proving up your case?
Speaker 6
The most challenging part was that this was a new technology, that no one knew how to get the data from it or know how to prove that the car was at fault. No one knew. This was all new. I wasn't even aware that these cars were on the roads, and they were driving themselves. The most challenging part was getting our hands on that data, And this is a years long it's going on seven years now.
And that's what really changed the case was an anonymous hacker that was able to get the autopilot computer and get it and see what the car was seeing and processing through the augmented video. And at first, officer Rizzo, Florida Highway Patrol, on June 19, took took the autopilot computer to Tesla headquarters, and Tesla gave it back to him saying there was nothing on this. It's just corrupted, and they gave him encrypted data that he couldn't even understand. You know, the that was incredibly hard of getting that data, and we finally got it. And the car and that data showed that the car recognized pedestrian, a stop sign that was 99% sure there's a stop sign at one point, and the vision system was confused.
It thought my vehicle that was parked off the road was a bicycle, a car, an SUV, a minivan, and it went straight to the stop sign, ran a red light, ran over intersection signs, and all the system decided to do was abort autopilot a second and a little bit more than a second before impact. You know, it's not it's not just to not fully report these crashes. There was, like, some order that was done where now they have to report any cars that are autonomous or on autopilot within five or ten seconds of a crash to prove if autopilot was at fault. It that was very hard. It's it was also very hard, you know, seeing Elon Musk on news stations in The US and even on his website, Tesla, where people buy cars, showing the car driving itself, stopping at stop signs, turning, navigating from point a to point b, and then come to the courtroom and say, no.
This is just your grandma's cruise control. It just keeps you centered in the lane. And, I mean, it's a bunch of they're they're hiding behind loopholes. This whole level two and level five I mean, this car is advertised as level five, and they're trying to be under level two. There needs to be change with that.
This was very it's an incredible battle that we're going against, going against Elon Musk and Tesla. And, you know, we're currently on the appeal process, and we're prepared to take this to the Supreme Court. Mhmm. And that's not that's not right for a family that lost a loved one and someone that was injured. And, you know, we're here to hold them accountable and to hopefully encourage the regulators to put the right rules in place.
Thank you.
Speaker 1
Yeah, thank you. I so an unfair question to ask, you know, someone who had to go through that Yes. To, you know, to now I'm going to ask you an expertise kind of question and it may very well be above your pay grade as they say. It's above mine. When we ask the question in the first place.
It's not a question I know the answer to. But what should be available when there's a catastrophic accident like that? A black box like they have in commercial air flight? Or what? Have you thought about that at all?
During the course of the trial, I'm not asking what your attorneys told you, that's not appropriate. But did you have thoughts yourself about if if only it were the law that x has to be available, this would this would have all been a lot different. What's x?
Speaker 6
I mean, firstly, these cars need to be fully tested before they're on the roads. You know, these they just threw this car on the road, and they were learning as they were going with crashes that were happening, training the computer to not do it again, and we were pretty much guinea pigs. You know, there should be standards where, automatically, the data, the vision system, what the car is processing and seeing, not just a bunch of logs, which those are also important, but we need to see what the car was seeing and processing. You know, the highway patrol, the police officers, the lawyers, the judges, insurance companies all should have access to this. This shouldn't be something where a lawyer or a police officer has to go to Tesla and ask them for it.
It should be something that automatically they can get on their own. There's there's so many layers layers to this, but I'm here for the fight.
Speaker 1
Yeah. Well, thank you. And, again, we appreciate your stamina and courage and resilience to come here today and testify. And before I thank all the panelists, I want to to Senator Valderas, who's joined us. It looks like she would like the floor.
Speaker 10
Thank you, Mr. Chair, and I appreciate this hearing on this very important issue. It's an emerging reality that we're faced here in the state of California, across the nation and the globe. My question is for Doctor. Missy Cummings, who's on the line.
Is she still on the line? Do we know?
Speaker 10
So we're kind of I feel like we're we're in between two places right now and that we need some policy parameters, guardrails that ensure that the public is safe. But, also, we don't wanna stifle technology. What's that what do you have any recommendations, any insight as to what the guardrails should look like could look like based on your research and expertise?
Speaker 4
No. I think it's a great question. I am not at all against self driving cars. I have a teenager who just crashed my own car, so I I want them more than anyone. But I also think that we are in this wild wild west of understanding that, look, self driving cars are not fully autonomous.
If they were, they wouldn't need human babysitters. I applaud the companies for using the remote operators. But instead of trying to pretend that they're not doing as much as they are, I think we need to embrace them. It's a human AI collaboration, has jobs that go with it. Great.
I I think that we need to be open, and states and the federal government needs to look at the remote operations and say, look. This is it's risk management. It's gonna be good for the companies. It'll help build public trust. So instead of thinking remote operations are bad, we need to think that remote operations are good.
But, also, what goes along with embracing remote operations is more transparent access to data. We need data in terms of how many operators the company's got, how often they're intervening. This is data that just simply doesn't exist right now even though Ed Markey, a senator from Massachusetts, has been asking for and the companies keep pushing back. So more data about what's happening with remote operations, but also remote more data about this phantom braking problem. I cannot emphasize that if self driving trucks are allowed to operate without a human in them, they are going to kill somebody.
This phantom breaking problem is very serious. But it's not fair to hold all companies to one standard, so we need to or to to yes. We should hold them one to standard, but we shouldn't point the finger at all of them unless we have their data. So we need hard breaking data. I know that your DMV is looking at this.
So more data, hard breaking data, remote operations data. Their crash data that the the companies are giving you, they like to spend the data their own way. Of course, academics, we like to spend the data in in our own way. But I think having the data having them give the data to third objective third parties to validate the results that so if the Waymo, for example, wants to say we are reducing accidents by 92%, great. Give that data to a third party for independent verification.
Speaker 10
And so would you like to elaborate? I think it was brought up earlier on this concept of a black box, such as airplanes are required by law to have. Can you elaborate on that at all, on your perspective of whether or not that would be beneficial, needed? And also, do you think that there is that self driving system should be required to have a backup, some type of any type of backup in the event its primary sensor suite fails?
Speaker 4
Well, this is complicated. Maybe we should have a workshop that I can come for an entire day and talk about it. I will tell you that this idea of a digital black box that could go on, not just self driving cars, but driving assist cars, this has been thought about it, and it's, at the federal level for some time. And, again, the federal government has slowed action. Maybe over the next few years, this will gain momentum as more accidents happen.
I will say that what happens with self driving cars are they are just gigantic sensors. There's no data that you can't get off of a of a self driving car right now that would have to come in some other form. Now whether it all goes to this one centralized repository that comes off with every accident, I think that's a a situation that needs to be discussed. And and, unfortunately, I think that this I agree with Ariel on this point. It really needs to come at the federal level because, you know, you don't wanna have different states having different regulations if we're gonna have like, for the reason that EDRs, these are electronic data reporters that have been so useful is because all states use them.
All cars have them. So I do think that that's something if we're going to go down that route, we need to kinda work together across state lines. That being said, every self driving car that is operating on California roads right now has all the data that you need to look at an accident, including video. They must and often they do, but I think for every accident, the the recordings, both internal and external, should be that the normal part of data that is turned over because what you will see in the video will often show a different story than what the companies are reporting in the federal reporting date.
Speaker 1
All right. And I do want to come back to Mr. Wolf and just ask I asked Cinder asked a question earlier. It should probably be self explanatory when he gives me the response to it, if not, I'll repeat. Thank you.
Speaker 3
Sure, Mr. Chairman. So I think the idea is a 92% improvement. What happens with the remainder and how do you set up guardrails with what should be done in those situations? I think it starts with the idea that in any situation, the facts will determine liability.
But I think beyond that, California's regulatory structure has a permitting structure, not all states have that. And so there's always the idea of rescinding permits. You know, Texas just updated its law in a certain respect for operations to have a process by which, if there is a situation that the authorities have determined there is a reason to try and make impound a vehicle or to make certain judgments about that, to engage in that process and remove the authority to be able to based on the representations the company has made to the state. So, think there are a number of analogs. Our association has gotten involved in our model legislation across the country and thinking about that, And certainly, we'd be happy to have further dialogue on how to how to do it.
Of course, the the last piece of the puzzle is at the federal level, anytime there's, some issue or incident of a safety nature, NISA can, and it does, now repeatedly, open those kinds of investigations.
Speaker 1
Yeah. Short of a appreciate that very much. Thank you for the thoughtful answer. Short of a recall or complete voiding of the permitting or licensing up to that point. I think the human nature is to go back to the system of motor vehicle regulation that we have had since Henry Ford and say, look, it's been regulated pretty well in terms of accountability.
Again, that was an accountability question. People get a ticket if they go too fast. People get more serious consequences under the law, not just in terms of civil liability. In many cases, we just had a big discussion in Senate Public Safety about that today and moved a bill along that will increase liability. None of that would be applicable to autonomous vehicles.
And maybe there's a difference in the fact that the fact patterns that you see aren't going to be the same with autonomous vehicles as they would be with a human driven vehicle, including the most obvious one, is you don't have a person there that can account for their actions behind the wheel. We've established today there's humans involved somewhere in oversight. But that's where we could use help. I'm not asking you for further response. I think if the industry and this is one person's, one senator's opinion if the industry wants to avoid a barrage of attempts to figure out ways to hold the industry accountable.
The best way to do that is collaboratively, is to come back and say that's why I used the term best practices earlier, look, this should be the penalty for blocking traffic in the middle of a street or whatever the case may be. We know what that penalty is to a human driver right now. I don't think, in fact, I know that in the state of California right now, we we don't have a remedy. And again, just to cite the old maximum of equity, we live in a world where we believe for every wrong there should be a remedy. And I think people are struggling right now because they're seeing wrongs or potential wrongs, even if there are eight cases out of 100, that there should be a remedy for.
And we have no place to go with that. I don't think that's healthy is what I'm saying. I'm not here to lecture you. I think sooner or later what you end up with is regulation foisted upon you, rather than regulation that's brought forward in a mutually beneficial way. So that would always be the hope.
Thank you all for being here, especially our long distance traveler and you're obviously extremely powerful testimony, I hope you're willing to keep doing that. I imagine there may be committees here in the future that are actually voting. This is an informational hearing, but they're actually voting on new policies and new bills that are going to want you back as a witness if we can get you back here in the future. And, Mr. O'Dowd, thank you very much for representing.
With that, we'll move on to the next panel. Thank you.
Speaker 1
The next panel in Thank you, Doctor. Cummings. The next panel is Adam Wood, San Francisco Fire Department, former executive board member of the San Francisco Firefighters Brandon Sanchez, deputy chief San Jose PD Joseph Augusto, California Gig Workers Union and Shane Guzman, Teamsters, California. Welcome to all of you. Looks like everyone is here and accounted for.
You can really go in whatever order you want, but why don't we start with Mr. Wood, since I called your name first. And again, five minutes. Hopefully, the rest of the way, the Q and A will go a little more briskly. That was on me and mostly the last panel, but I had a lot of questions.
It is what it is. But I appreciate you trying to keep to about five minutes on your opening, and that'll give us more time. Thank you.
Speaker 9
Hi. Good afternoon, Chair Cortese and members of the committee. My name is Adam Wood. I retired from the San Francisco Fire Department in June 2024 with twenty nine years of service, and I served on the executive committee of the firefighters union in San Francisco, Local seven ninety eight, beginning in 2012, and I'm here representing that organization. I'm speaking to you today about our experience in San Francisco with autonomous passenger vehicles, also known as robotaxis, and the challenges that they've presented to firefighters and other emergency workers in the city.
There's been a number of issues, but the main problem that we've had with these vehicles as first responders is their tendency to shut down and immobilize themselves when confronted with an emergency scene or with emergency vehicles responding to a scene with lights and sirens. We've had AVs pull into the middle of an active shooter scene and shut down, thankfully without a passenger in that case. We've had them drive into a fire scene and park on top of a dry fire hose. That prevents that hose from being charged with water and used to extinguish the fire. We've had them block in an ambulance with a patient on board who needed to be transported to the hospital.
We had one shut down directly in front of an apparatus door at a firehouse. As the apparatus door was opening and a vehicle was preparing to leave for an emergency response, taking that vehicle completely out of the emergency response. We've had multiple cases on the narrow congested streets of San Francisco of an autonomous vehicle confronted with a responding emergency vehicle shutting down, blocking all traffic, and making that route unpassable. And last December 2025, we had a massive power outage affecting the western side of the city. Traffic signals went out and we lost local cellular coverage.
And the result was we had dozens of autonomous vehicles immobilized at the intersections throughout stopping traffic throughout entire neighborhoods from anywhere between twenty minutes to almost an hour. So the solution that the AV companies have presented to us as a city is they provided the city with a hotline that allows dispatchers to contact a remote operating center and have a remote operator move the vehicle. As you may expect from your own experiences with remote services, it's not been an effective solution. Often, the dispatchers have trouble reaching the remote operating center, and once they do, the remote operators, in some cases, can still not move the vehicle. In the power outage in December, the city reported 31 calls from dispatchers to the hotline, with many of them put on hold for extended periods of time.
One dispatcher was put on hold for fifty three minutes. So ever since San Francisco became the California laboratory for autonomous vehicles, the San Francisco Fire Department leadership, both on the labor side and the command staff, have appealed to the autonomous vehicle companies and the CPUC to, first of all, slow down the deployment so that we could address these problems we were beginning to see in the field. And also, one of the key things we were asking for was a public safety manual override option that would allow a police officer, a firefighter, or a paramedic confronted with a stalled vehicle at an emergency scene or en route to move it without the intervention of a third party, a remote operator. We had hoped that what we considered to be a common sense solution to the problem would be agreed to voluntarily by the companies. But the years have gone by and here we are still seeking help to help us weigh in on crafting this technology so that it serves its purpose but still serves the best needs of the citizens we're trying to protect.
I believe it is possible that we can come to a solution that gives us effective autonomous vehicles with no negative impact on public safety. But at this point in time, I don't think it's going to be possible without some element of the tool of local regulation so that we can weigh in and actually have an impact on what's done with this technology. Because committee members, December last past December was a wake up call. In San Francisco, we are going to have a disaster, whether it's an earthquake or whether it's a windstorm fired followed by a conflagration such as Los Angeles experienced a little over a year ago, where we're going to have a combination of a power outage and the need to rapidly evacuate residents from danger and create routes for emergency workers to respond. If we have a fleet of stalled emergency vehicles blocking those routes, there's going to be serious consequences.
So there is some urgency to this need for reform. In the future, anything you, as senators, can do to help us will be greatly appreciated. Thank you for your time, and happy to answer any questions.
Speaker 1
Thank you. We'll get to questions at the end of the panel. Who is next on the agenda? Brendan Sanchez, San Jose Police Department.
Speaker 11
Good afternoon, Chair and members of the committee. Thank you for the opportunity to speak to you today regarding autonomous vehicles and public safety. My name is Brandon Sanchez. I'm the deputy chief for the San Jose Police Department, and I'm be I'm I'm here on behalf of the California Police Chiefs Association representing municipal police chiefs and public safety leaders across California. Autonomous vehicle technology represents one of the most significant changes to transportation in decades.
From a law enforcement perspective, our priority is simple, safer roads and safer communities. We recognize that autonomous vehicles have the potential to reduce crashes, particularly those caused by impaired, distracted, or fatigued drivers. Because human error is leading is the leading cause of collisions, this technology offers real promise to improve road roadway safety. We are also beginning to see how autonomous systems can support public safety operations. Currently, the Miami Dade Sheriff's Office is piloting a the first in the nation autonomous law enforcement support vehicle equipped with tools such as deployable drones and real time situational awareness capabilities to assist officers in the field.
These technologies could help improve response times, expand officer awareness, enhance safety for both first responders and the public. California police chiefs recognize these benefits and support reasonable innovation. At the same time, the technology is still developing and presents real world challenges. Autonomous vehicles today are not fully independent and do not always respond predictably in complex environments. Officers across California are already encountering issues in the field.
In Santa Monica, a Waymo vehicle struck a child near an elementary school prompting a federal investigation. Autonomous vehicles have blocked fire engines, ambulances, and police vehicles responding to emergencies. There have been incidents where vehicles stopped in active traffic lanes and failed to navigate around emergency vehicles and emergency scenes. These are not theoretical concerns. They are real operational issues affecting public safety today.
One of the biggest challenges for law enforcement is the lack of clarity around responsibility and enforcement. When there is no driver, it is it is unclear who is responsible for a traffic violation, who receives a citation, and how existing laws apply. Officers are still adapting to training around how to properly redirect these vehicles when necessary. Without clear authority and adequate training, enforcement becomes difficult and inconsistent. It is critical that autonomous vehicles can safely interact with first responders.
Officers must have a clear and reliable way to stop a vehicle, redirect it, ensure it does not interfere with emergency operations. While ultimately an officer can rely on their authority to remove an an obstruction, the delay of even a few seconds at an emergency scene can have serious consequences. From the perspective of California police chiefs, several principles are essential moving forward. Public safety must remain the top priority in all deployment decisions. Law enforcement must be consulted at every stage of deployment and implementation.
This includes at the local level where more needs need to be done. There must be clear statutory authority for enforcement and accountability and rapid responses required of the vehicles during an emergency. Standardized protocols must exist for first responders' interaction with autonomous vehicles, and training is essential. Transparency is critical when incidents occur. We wanna be clear.
Law enforcement is not opposed to autonomous vehicles. We see the potential to reduce crashes and save lives, but innovation must be paired with accountability and safety. The margins on roadway safety and in in in legislative decision making are often very small. A handful of incidents or a few seconds of confusion in an emergency can have life altering consequences. California has the opportunity to lead in this space, but we must do so thoughtfully and responsibly.
Thank you for allowing me to share my thoughts, and I'll be here to answer any questions.
Speaker 12
Good afternoon, Chair and members of the committee. My name is Joseph Augusto. I am a San Francisco resident and have been a full time Uber and Lyft driver in the Bay Area for the past decade. Over that time, I've completed more than 25,000 rides and maintained consistently high ratings. I spend the majority of my time and waking hours on the road.
I take pride in operating safely and professionally. I'm here today to express my serious concerns about the safety of autonomous vehicle passenger services. In San Francisco, San Francisco is the center of this issue. The Bay Area is the birthplace of robotaxi technology, and currently, more robotaxis there are more robotaxis there than anywhere else in the country, making it a real world testing ground for this technology at scale. Over the past two years, there have been a noticeable shift in the autonomous vehicles' behavior.
Early on, they tended to be overly cautious sometimes to the point of outright slowing down traffic. But more recently, I've seen the opposite. I now frequently see autonomous vehicles accelerating aggressively, creeping onto sidewalks, and making maneuvers that create confusion and risk for everyone around them. I have personally experienced two near collisions involving autonomous vehicles in San Francisco. One occurred near Dubose Triangle and the other at the intersection of Noe And Market Street.
In both cases, the autonomous vehicle ran a red light. As a professional and experienced driver, I am able to anticipate hazards, but these incidents were abrupt and unexpected and dangerous. Safety on our public roads must always be the top priority. For those of us who drive for a living, the margin of error is small and the consequences of mistakes can be life changing. What concerns me even more is that my experience sharing the road with autonomous vehicle is not isolated.
Each day, there seems to be more reports and video evidence showing autonomous vehicles violating basic traffic laws and creating unsafe conditions. For example, the following reports, and some have alluded to some
Speaker 12
things already. An autonomous vehicle carrying passengers blocked emergency responders during a mass shooting incident in Austin. Also in Austin, there have been more than 20 incidents of robotaxis illegally passing stopped school buses. Another autonomous vehicle drove through a police standoff in Los Angeles with a passenger inside. In Chandler, Virginia, an autonomous vehicle stopped in front of ongoing traffic to make a left hand turn.
I'm sorry. That's not funny. But in Santa Monica, an autonomous vehicle struck a child near an elementary school. These incidents point to a broader pattern. When these systems fail, they can fail in ways that are unpredictable and difficult for human drivers and first responders to manage.
In San Francisco, we have seen additional challenges. Many of you are aware of the robotaxi disruption during the December power outages last year. Hundreds of vehicles stalled on the public streets creating traffic congestion and safety hazards. I experienced this firsthand. I was stuck in traffic behind stalled robotaxis unable to move for an extended period of time.
Our board of supervisors, transportation and land use committee held a informational hearing to get to the bottom of what happened. We came out of that hearing what came out of the hearing was more even more disturbing concerning. During those power outages, city officials in charge of emergency response responses reported that there were significant delays in connecting with the companies responsible for stalled autonomous vehicles, and that first responders were forced to physically move disabled autonomous cars out of the roadway. That raises serious questions about preparedness and accountability. If these vehicles depend on remote assistance to function properly, then they then that support must be reliable, immediate, and adequately staffed, especially during emergencies.
When it's not, the burden and the cost shift to our cities or first responders and the public. I believe autonomous vehicles must be held to the same standards as human drivers, and in many cases, to even higher standards given that they are being deployed to scale on our public roads. First, robotaxi companies should be required to adequately staff and support their fleets so that they are not relying on emergency responders or our city personnel to fix malfunctions. Second, local authorities must have a clear tools to hold these company companies accountable. That includes the ability to to issue fines and in cases where repeated violations to suspend permits until safety compliance can be demonstrated.
Third, the public deserves transparency. We should have access to timely accurate data about permits, safety records, incident reports, and traffic violations. Without transparency, there is no accountability. And without accountability, there is no trust. I urge you to act now to ensure that our streets remain safe, not just for passengers in these autonomous vehicles, but for pedestrians, cyclists, first responders, and professional drivers like me who share the road every day.
Thank you for the opportunity to testify. I look forward to your leadership on this critical issue.
Speaker 2
Good afternoon, mister chairman, members of the committee. Thank you for inviting me here to speak on this important topic. My name is Shane Gusman. I am the legislative director for Teamsters California, which is the statewide organization for the international brother of Teamsters. Spent the last three decades representing the Teamsters here at the state capital, among other things, on highway safety issues.
And that's why I'm here today to speak to you. We are here because California is at a crossroads in our view. The Department of Motor Vehicles is taking us down a dangerous path. The DMV's proposed heavy duty autonomous vehicle regulations are not cautious. They are not incremental.
They are not safety first. They are a green light for the rapid and reckless deployment of heavy duty autonomous trucks on California's public roads. And that should concern every single one of us here. Let me be clear about what is happening. Up until recently, California law recognized a basic reality.
Vehicles weighing over 10,000 pounds, large commercial trucks pose uniquely high risks, and they should not operate autonomously without strict safeguards. These regulations or proposed regulations erase that line. They would allow autonomous trucks weighing 80,000 pounds to operate on public roads, not after proven safety, but based largely on manufacturer self certification. That is not oversight in our opinion, that is abdication of responsibility. And the risks here are not theoretical.
When something goes wrong with a passenger vehicle, as you've heard here today, the consequences are very serious. However, when something goes wrong with an 80,000 pound truck traveling at highway speeds, the consequences can be catastrophic. Stopping distances are longer, kinetic energy is exponentially higher, the margin for error disappears. And yet the DMV is proposing to allow these vehicles onto our roads without requiring independent safety validation, without meaningful operational limits, and without demonstrated performance in California's real world conditions. California is not Arizona.
It is not Texas. Our roads are more complex. Our traffic is denser. Our conditions are more extreme. We have dense urban corridors filled with pedestrians and cyclists.
We have mountain passes, winding roads, steep grades. We have fog, wildfire smoke, flooding and snow. We have some of the most complex freeway interchanges in the world. And yet, these regulations allow companies to prove safety in other states and then deploy here. That is not a safety standard.
That is a loophole. Even more troubling is how these regulations define safety. Instead of requiring independent third party validation, the DMV relies on something called safety case. A safety case, excuse me, a document prepared by the manufacturer itself. No independent audit, no external verification, no enforceable performance threshold.
We are being asked to trust that companies will police themselves when human stakes are alive human lives are at stake. That is not how we regulate companies that put dangerous products on the road, and that is not into the stream of commerce, and that is not how we should regulate these vehicles. The proposal also fails to place meaningful limits on where these vehicles can operate. There is nothing that clearly prevents deployment in residential neighborhoods, near schools, or in dense urban corridors, precisely environments where the technology is least reliable and where the consequences of failure are highest. And finally, these regulations allow deployment far too early.
A company can move from testing to full deployment without demonstrating consistent real world safety performance across diverse conditions. There is no clear minimum performance standard, no staged rollout tied to proven safety benchmarks, no requirement to earn public trust before scaling. That is not careful governance. It's a rush to deployment. At its core, this raises the fundamental question, what is the role of government here?
Is it to facilitate the rapid commercialization of unproven technology to benefit a few companies? Or is it to protect the public? Because right now the DMV is choosing speed over safety. We are not opposed to innovation. But innovation cannot come at the expense of public safety.
And it certainly cannot come at the expense of basic regulatory responsibilities. If the DMV will not step up and fulfill its duty to protect the public, then this legislature must. We cannot stand by while heavy duty autonomous vehicles and all autonomous vehicles for that matter are deployed under conditions that endanger our communities. If the regulatory agency will not put safety first, it is this legislature responsibility to ensure that no AV deployment proceeds until public safety is truly guaranteed. Thank you.
Speaker 1
Thank you. Thank you all. We'll turn to the senator to see if he has any questions or comments to start.
Speaker 13
No, I think you guys covered it very well. All of the public safety aspects of the autonomous vehicles. I watched them perform in Santa Monica and watched them not perform in Santa Monica before. And so, you know, I think it's technology that's going to continue to emerge, we do have to figure out that sweet spot of, you know, hey, where they're appropriate and where they're not, where they don't work the most because, you know, I can see some advantages for some people that are not able to drive a car themselves and still can get around town using one of these vehicles. But when they're in condensed areas, it's just like anywhere else.
Sometimes they become their own liability, and we have to kind of sort through that. And as the technology is helped along to see where it can go, we also have to provide some backstops. And so I think that's what people are asking here is looking at backstops. That's all.
Speaker 1
All right. Thank you, Senator. The question for Mr. Wood, I guess that's your title as retired. Response times and I don't know how long you've been out of that loop, but in the San Jose area, both as a county supervisor and as a council member, I was quite aware of statutory or contractual response times that were set up for EMTs, particularly coming out of either fire or private ambulance companies.
Do you recall what those would typically be in San Francisco? And I'll give you, deputy chief a chance to respond to if you know. I didn't come here knowing the answer, but I'm curious about how much time would be adequate response time for somebody responding from an autonomous vehicle company to a scene?
Speaker 9
Yeah. For cities that have two zero one rights to provide their own ambulance service, San Jose also and San Francisco, there are state regulations on response time for those units, for those medical units. And San Francisco was making steady progress towards lowering those response times. And then we have seen a spike again. And it's not that every delayed response is because of a direct interaction between an autonomous vehicle and an ambulance.
But I will say that in San Francisco that the volume of autonomous vehicles on the road has added to congestion in noticeable ways. And the one of the problems we have is we cannot get a direct answer from the companies on what their actual fleet sizes are. It's just what we observe. And that is slowing response times across the board, especially, I'll say especially in the central and northeastern part of the cities, which are the most congested part of the cities. We've seen a noticeable spike again where we were always hovering right on the edge of the limit required by the two zero one rights.
We occasionally go over it the allowable percentage of the time, but the autonomous vehicle congestion has definitely made the problem worse.
Speaker 1
Yeah, I appreciate that. And I didn't want to get into, you know, who's which city is meeting their response times, which isn't. I was more looking for a standard. I guess really, though, in the context of this, what happens if you meet your response time and an autonomous vehicle is a part of the collision at the scene?
Speaker 1
really asking you the question. How soon do they need to be there? If if you were in in in your mayor's shoes or my shoes, actually, the mayor can't promulgate any laws because state law preempts San Francisco and San Jose from doing any local control. You alluded to that earlier. But if you could, what would you be trying to align?
Is there an alignment in terms of response time? I'm really thinking in terms of passengers, could be no fault of Waymo or any the other companies, passengers trapped in a vehicle.
Speaker 9
Well, that's Soon as that need to happen. We we would need to have it. I mean, if once we're on scene, we would need that to happen in less than five minutes. You know, so it's not so much the response time to the scene, it's trying to get access to the patient on scene if it's create there's some obstacle in
Speaker 9
Yeah, would have which is again why we were the idea that came to us was the idea of a public safety manual override so that we could take, whether with a code or a key fob or something that doesn't require contacting a remote operating center, being able to take control of the vehicle, either to make sure the doors are unlocked or to move it out of harm's way, and, yeah, in that case, to access a patient. So we would need it in a much shorter time than we're able to do currently using a hotline to the remote operating center.
Speaker 1
My recollection is that the Santa Clara CountySan Jose contract response time was somewhere around eight and a half. Yeah. That's In other words, if you don't meet it, you don't get paid, basically.
Speaker 9
Right, right. That's right. You don't meet it, you don't get paid, and you can lose your ability to provide a municipal ambulance service if you're missing those numbers consistently.
Speaker 1
Okay. But you're saying those are that information should be readily available at the state? Oh yeah. Through the 201? Certainly.
Okay. Yeah. How does dispatch, you used one example of the many that you gave about a dispatch delay. And I don't know I I also, having been in county government and city government, I understand there's people who do nothing but the ins and outs of dispatch. So if you don't know the answer, that's fine.
But what I'm trying to understand is how does dispatch work now under the current scheme of things with an autonomous vehicle situation? Because you alluded to a delay of fifty six minutes or something like that.
Speaker 9
Right. Those that was so that we have a twenty four hour dispatch center, emergency command center, who are it's a combination of police, fire, and medical, all coming from a central location. So those dispatchers are sending out the appropriate units to whatever the 911 call responds to. And if the responding units encounter a delay caused by an autonomous vehicle, they'll communicate that to the dispatcher and the dispatcher will utilize the hotline to try to get the remote operating center to move the car if that's what the problem requires. But that's where we're running into these delays.
Speaker 1
So the autonomous vehicle companies, at least as of now, are not affirmatively notifying dispatch that their vehicle has been in an incident?
Speaker 9
No. When these calls to the hotline are being made, they're usually requested by units in the field.
Speaker 1
So you're already late? You're already late. Okay. Yeah. And is my understanding is there's technology out there, and it may be just my experience with my own non autonomous vehicle, my EV, with OnStar systems and such that you're in an accident and immediately there's a dispatch notification that comes from that vehicle.
I don't know if you're aware of that. Is that technology at least generally available through municipal systems? I don't mean provided by San Francisco, but is it compatible with San Francisco's dispatch?
Speaker 9
It may be the overwhelming majority of the problems we're running into is not involving an AV in an accident. So that it might not trigger that on star type response because it's not experiencing an accident itself or become inoperable because of a of a trauma or a but it's it's just frozen by an unusual situation. Now, so I don't know if that triggers any response from the vehicle to its own remote operator, but it certainly seems that the call has to come from us before anything actually happens.
Speaker 1
Would it be helpful to have the technology in the autonomous vehicle notify its remote operator and the remote operator then notify dispatch?
Speaker 9
If that would speed up the resolution of the situation, it would be very helpful.
Speaker 1
Okay. Any I don't know if you're aware of the technology around dispatch in San Jose. I know it's a good system. We've San Jose guy, but what do you know about that? Yes, thank you, Senator,
Speaker 11
for the question. We have interacted with OnStar and similar technology when we've had instances of crashes involving the vehicle that has that capability in it. They'll contact our dispatch. Our dispatch will pass them on to the officers in the field. But to my understanding, in the autonomous world, we have to generally, like my colleague said, make the first call into them first.
We don't get anything from them on the front end.
Speaker 1
Right. Okay. If you learn anything more about that, the aftermath of this hearing, in fairness to all involved, please let us know. We'll certainly distribute that information to committee members, folks who are, again, later on dealing with bills and policy and trying to maybe fix some of the rules. Yes, Senator Seyre?
Speaker 13
I had a question on you were talking about the response time. Yes. Were you talking about from the time that they dispatch to the time they arrive on scene? Yeah. Because the the response time that is recorded and and for contracts for ambulance companies and stuff is is that time.
Once you get on scene, you just punch on scene. And then you try to figure out what needs to be done. In the case where they have patient inside a car and they can't get access and they can't get ahold of Well, then the car becomes a demolition vehicle and they start cutting it up and you you make forcible entry. So that's not as I think it's not as big as the issue. If there's an an autonomous vehicle blockage and they can't get through to a call, then the dispatchers will will dispatch the next closest rig and they talk to each other about, hey, you're going to have to come from the other side and get to whatever address we're going to.
So there could be a potential delay in that process. But as far as getting on scene, it's whatever it takes to get from the And it's usually Those contracts are usually five to ten minutes depending upon where they're located And if they're out in the country or if they're out in a city or wherever it is, that's where that comes in handy.
Speaker 1
Yeah, thank you. It sounds like as of given the experience you have as a high level first responder, it reminds me of the days I sat. I had a seat mate, John Dequisto, the late John Dequisto, who was a battalion chief in the day, retired, served on the city council and we were having a similar conversation about fire apparatus access when blocked. He said, he leaned over and said it won't really matter because they'll just take the truck and move everything out of the way. But in the issue of an injury, we heard testimony earlier of an injury more than we've heard testimony from all of you of certain injuries that have occurred.
And clearly, can happen again. My concern, I'm just trying to get at the issue of as the legislature is trying to work with the industry on what would be an appropriate response time standard that aligns with public safety standards, what would that be? And I think I have a general answer to the question. It sounds like it might be in the eight minute range. You made it very clear we can find that out right here at the state of California, so we'll go do that.
Anything else, senators? If not, I'm going to let this panel go. We appreciate all of you being here. Appreciate your testimony very much. It's added greatly to the hearing today and informing a
Speaker 1
of people. I should mention, I think all the Sacramento folks here know this, but we've had people coming in and out. Of course, they're on other committees and multiple committees meeting today. In fact, I keep checking my text here to see when they're going call me over to vote in the Audit Committee. Any time now, I'm going have to leave for a few minutes.
But just know that much like City Hall and so forth, this place is completely wired. We have staff and all the floors here watching on monitors, listening, members listening, and people, of course, who are gonna run back to tape afterwards and utilize your testimony, hopefully, for good things to happen. So, thank you. The next and final panel on state regulatory agencies, I want to invite Bernard Soriano, Deputy Director of Policy, Department of Motor Vehicles Miguel Acosta, Autonomous Vehicles Branch Chief, Department of Motor Vehicles and Tara Curtis, Director, Consumer Protection and Enforcement Division, California Public Utilities Commission. Thank you all for being here.
Welcome. And I'm gonna go ahead and get you started. I see someone waving me, which probably means that I have to run over across the street and run back again. So I'll put one of the senators here in charge while I'm gone, but we'll get you started in five minutes each. And Mr.
Soriano, since you were listed first, we'll call on you first. Thank you.
Speaker 14
Well, good afternoon, Mr. Chair and committee members. Thank you for inviting us to testify. I'm Bernard Soriano. I'm the Deputy Director at the DMV.
And joining me at the table is Miguel Acosta. He's the chief of our autonomous vehicles branch. Miguel will be reading a prepared statement. But before doing so, I wanted to point out that California has started regulating AVs since 2014. Well over a decade.
And those regulations have evolved to the point where today California has a coordinated end to end framework for vehicle operation and passenger service, with active oversight and enforcement at every stage. Thank you, again, for inviting us and giving us the opportunity to testify. Miguel will go ahead and read the statement. And, after that, we will be available to address any questions that you may have.
Speaker 15
Good afternoon, Chair, Committee members. As Bernard mentioned, my name is Miguel Acosta. I'm chief of autonomous vehicles here at the California DMV. Senate Bill twelve ninety eight in 2012 established vehicle code section thirty eight seven fifty, directing the DMV to adopt regulations necessary to ensure the safe operation of autonomous vehicles on California public roads. Since then, the department has adopted three rulemakings authorizing testing with a safety driver in 2014, driverless testing and deployment in 2018, and light duty autonomous motor trucks in 2019.
The department is now developing a fourth rulemaking package. Throughout this process, we have engaged extensively with local partners, including transportation agencies and first responder departments in cities like Los Angeles, Santa Monica, and San Francisco, and we've convened two statewide first responder roundtables to ensure operational concerns are fully incorporated. The new proposal establishes a path for testing and deploying heavy duty autonomous vehicles and strengthens the department's safety oversight. It adds more detailed and frequent reporting including vehicle immobilizations, system failures, hard braking, and aligned federal crash reporting requirements, and expands the department's enforcement tools. It also implements Assembly Bill seventeen seventy seven by requiring AVs to respond appropriately to emergency geofence messages, comply with law enforcement direction, and support first responders when an override system is present.
The rulemaking also creates a formal process for law enforcement to notify the department and manufacturers when an AV is involved in a moving violation. California's approach continues to align with the federal model, where manufacturers self certify compliance with federal motor vehicle safety standards or receive exemptions through NHTSA while the state oversees driver licensing, vehicle registration, insurance, and operational safety. The department currently administers three AV permit types: testing with a safety driver, driverless testing, and deployment. Today, 28 manufacturers hold driver testing permits, six hold driverless testing permits, and three are authorized for deployment. Under existing regulations for testing with a safety driver, manufacturers must operate light duty vehicles that meet federal standards or hold a NHTSA exemption.
They must maintain $5,000,000 in insurance, report collisions within ten days, submit annual disengagement reports, and ensure their test drivers have clean driving records. Driverless testing carries similar requirements but requires a defined operational design domain, meaning speed, roadway type, weather constraints, time of day operations, a communication link to a remote operator, and a law enforcement interaction plan. Deployment builds on these standards by allowing manufacturers to receive compensation and requiring consumer education for vehicles sold to the public, as well as continued communication capabilities and an ODD description that reflects actual testing performance. The department maintains strong oversight through ongoing incident review, data reporting, and post permitting enforcement. DMV has issued three revocations and 14 suspensions since 2014, such as the 2023 suspension of Cruise's driverless testing and deployment permit in San Francisco, and the 2021 suspension of Pony dot AI's driverless testing permit following a system failure related crash in Fremont.
California's regulatory framework supports the continued development of autonomous technology while prioritizing roadway safety, compliance with state law, and safe interactions with other road users. We appreciate the opportunity today and welcome any questions you may have.
Speaker 13
Thank you very much for your testimony. We'll move right on to you, Ms. Curtis, and hear what you have to say. And then depending upon when the chair gets back, he may have some questions for you. But if he doesn't, we're going to go right to public testimony, and then we'll come back and answer questions afterwards, okay?
Thanks.
Speaker 16
You. Thank you, Chair and members of the committee. My name is Tara Curtis. I'm the Director of the Consumer Protection and Enforcement Division at the California Public Utilities Commission. And I appreciate the opportunity to speak today about the Commission's role in regulating autonomous passenger service in California.
The Commission's authority focuses on passenger service, ensuring that companies can safely transport members of the public for hire. We do not regulate all autonomous vehicle activity. Our responsibility is limited to passenger transportation, including ride hail services, as they're commonly referred to in the public. Private use, delivery services, and other non passenger operations are outside of our jurisdiction. Passenger safety is our foremost priority and shapes how we design and implement the permitting framework.
Under the Commission's permitting framework, companies must meet the requirements that apply to transportation charter party carriers, such as limousines and tour buses, in addition to autonomous vehicle specific requirements. Autonomous vehicle companies that seek to offer passenger service may apply for either pilot or deployment authorization. In both contexts, companies may apply for permits to operate with or without drivers. And commercial passenger service is only permitted once an operator satisfies the Commission's deployment requirements. Safety oversight does not end once a permit is issued.
Operators must continue to demonstrate that they can safely transport members of the public and comply with all applicable rules. As part of this ongoing oversight, permit holders are required to submit safety plans and operational data and comply with all reporting requirements. The Commission monitors operations, reviews complaints, and follows up when safety concerns arise. We may request additional information, conduct inspections, and take enforcement actions when necessary. When concerns are identified, staff may open investigations and issue citations or notices to cease and desist or other enforcement orders.
The Commission may also initiate formal enforcement proceedings, which can result in corrective actions, financial penalties, or permit suspension or revocation. These enforcement tools enable the Commission to ensure compliance with permit conditions, reporting requirements, and passenger safety obligations. We also coordinate closely with other regulators where our responsibilities overlap, including federal and state agencies such as the Department of Motor Vehicles. While the DMV oversees vehicle safety and operation on public roads, the Commission's role focuses on passenger service. This coordination ensures that safety issues are addressed comprehensively.
As autonomous vehicle passenger service continues to evolve, the Commission is actively updating its regulatory framework through rulemaking opened in August 2025 following the conclusion of the prior proceeding. Given how quickly the industry has changed, this rulemaking is focused on ensuring the framework keeps pace by strengthening safety requirements, clarifying accountability, and improving transparency in data reporting. Ultimately, our goal is that autonomous vehicle passenger service operates safely under a clear and enforceable framework. Our role is not to endorse technologies or pick winners, but to ensure that when companies transport members of the public, they do so under strong oversight and that we can act if problems arise. Thank you.
I'm happy to answer any questions.
Speaker 13
Great. Thank you. Senator Araqueen, do have any questions? Okay. So I would imagine the chair will have some questions for you.
One of the questions I have is, you know, it seems like the structure of accountability is in place. And maybe the breakdown is sometimes what happens out in the field, which is normally the case. We are hearing some testimony about delayed responses, delayed time when they try to call the company and they can't get ahold of them or they're put on hold. How are those issues handled by whatever department is notified?
Speaker 15
Yeah, that's a great question. So, as I mentioned, our framework includes a robust incident review. If we have incidents where, for example, there's a report of some blockage where, let's say, blocking a fire engine or police or those kinds of things, we meet with the manufacturer, try to understand that situation, fully understand all the facts associated with it, and in some cases ask the manufacturer how they're going to remediate that situation. Every type of situation is different. There may be a reason for that particular situation.
But if the manufacturer cannot remediate that particular situation, that is when our enforcement tools
Speaker 13
up. Great. And for each of you, how have we had a lot of incidents where you have been getting complaints from the various public safety agencies or whatever regarding incidents with autonomous vehicles?
Speaker 15
Yes, we've developed great relationships with locals. We meet frequently with LA DOT, San Francisco, Santa Monica. You know, for a period of time, we were meeting very frequently with SF Fire to understand some of these incidents. We would actually obtain they had reporting of these types of situations that we would review. They'd send them to us.
We'd review them. We'd talk with the fire department to understand them better, and then we'd meet with the actual manufacturer and go over those particular situations. So we had a pipeline in place to hear from both parties and to try to make sure that these issues are being resolved.
Speaker 16
And I would just add, as well, I don't have exact figures on those numbers of complaints or outreach we have, but what I would say is that we have a process like the DMV to review every single complaint that a member of the public reports to the commission, and that is a contact information that's available to the public on our website. In addition, we when we see incidents reported to us or through news media or other means, those are also issues that we follow-up on. We inquire for more information from the providers. So, I'm sorry, I don't have the exact
Speaker 13
classify the volume as light, moderate, or all the time? In other words, you've to hire three people to answer all the calls.
Speaker 16
Yeah. Well, I do appreciate that, in the past, we have requested staff resources related to our oversight of AV and been granted one or two analysts over the past six years, I believe. So that has happened concurrently with the growth in the industry. And so it's not an overwhelming amount for our staff capacity at this time. Thank you.
Speaker 13
Okay. Alright. At this time, what we'll do is we'll take public testimony. If you guys wanna line up at the microphone there and get your all of your comments organized in your mind into about a one minute sound bite, that'd be great. We're gonna try and get you out of here by 04:00 ish.
Hopefully, the chair will get back here by then. If not, you guys are trapped until he gets back. Go ahead with your comments. Got oh, try to take about a minute or so. You don't have to take a whole minute.
I just want make sure you knew that.
Speaker 17
Afternoon, committee. Mark Vuksevich from Streets for All. I want to first and foremost appreciate this hearing for happening. I think this is an important topic. I also want to stress that we do not have an adopted AV policy framework, but there are some things that we do believe, and I want to share those.
We do believe that AVs as a whole are very safe and could potentially bring a new age of street safety. We also believe that companies like Tesla, are in the ADAS level two system, have flagrantly lied to the consumer, and that has led to death and destruction on our roadway. We think that agencies like DMV and CPUC are not necessarily adequate for regulating these, the topics of AVs and building the public trust that these topics deserve. We think AVs are not necessarily the future of cities because we think, geometrically, cities still require public transportation, walking and biking, but we think they could be more of the future for suburbs and suburban transportation. Lastly, there are substantial questions that arise when it comes to questions of parking, congestion, and land use.
If I go to a city and want to go get ice cream and tell my AV to go circle the block for 20 times while I go get cream, what does that do to our downtowns and to our cities and to our urban places when everyone is doing that? I think that leads to questions of congestion pricing and parking pricing to deal with the negative externalities. Appreciate the committee for taking the time. Thank you.
Speaker 18
Hi, everybody. My name is Janice Jackson, and I thank you guys for listening to our concerns. Once again, my name is Janice Jackson. I'm a member of the California GIG Workers Union. I live in Sacramento, and I've been driving for ten years.
I'm here because of my concern about AVs. AVs haven't been yet started in Sacramento, but I see them on our streets and now soon to start operation. And what concern what concerns me is all that I have heard from other drivers from the Bay Area and how they have ran lights, blocked traffic, and caused accidents and hitting animals. I haven't experienced any of this. But AV companies need to be held to a higher responsibility for all of the things that I've just mentioned.
Speaker 18
drivers, we are held responsible for those, so they should be as well. Thank you.
Speaker 19
Next speaker. Good afternoon. I'm Peter LaRome Munoz with the Bay Area Council. We are here today to strongly support the continued safe deployment of autonomous vehicles as a critical pillar of our state's economic and transportation infrastructure. And this innovation comes with greater safety for drivers, pedestrians, and others along our streets and roads.
AVs are the only vehicles on the road that provide 100% data transparency. And in light of the committee's rightful recent attention to traffic fatalities, Golden State companies are stepping in to lead the way on solutions to road safety. While the state struggles with repeat offenders on the road, the AV industry is continuing to provide a safer alternative for Californians to get around. But if we move to a patchwork regulatory environment, we risk a fragmented system that isn't just a hurdle, it's a de facto ban. We believe the DMV and CPUC are the appropriate bodies to maintain rigorous, uniform safety standards.
We need one rule of the road for the entire state to ensure this technology can scale, operate reliably, and safely serve our residents. It is also important to differentiate AVs from the driver assist industry. There is a substantial difference in safety performance, operations, legal liability, and insurance between SAE Levels II or III vehicles and fully autonomous driverless vehicles. Level IV, or fully driverless vehicles, are very different from the operations, technology, and functionality of driver assist vehicles. Policymakers should not conflate the performance of one with the other.
Can you wrap it
Speaker 19
Just to remain a global leader in innovation, California must choose a unified statewide path forward that prioritizes both safety and growth. Thank you. Thank you very much.
Speaker 13
Next speaker. You have ten seconds. He used most of no, I'm just kidding. Go do what you need.
Speaker 20
Afternoon. My name is Vikash Shankar, and I'm a member of the California Gig Workers Union. I'm a resident of Fresno, and I've been driving for Uber and Lyft for the past ten years on and off. I'm here today because I'm concerned about how AVs are gonna affect my community and my fellow drivers even though they're not there yet. As a driver in the Central Valley and also as a former journalist, I've seen firsthand the impact of red light crashes and unsafe driving in our communities, virtually none of which were the fault of the professional, safe, and sober drivers that AVs are actually seeking to replace.
As a as a driver, there are already too many, makeshift memorials, on our busy intersections, too many families affected by inattentive and unsafe driving. I mean, even one of our intersections has a YouTube channel dedicated to footage of red light crashes and and and, you know, people running red lights. AVs are coming to our communities. We know this. And as a driver who takes pride in getting residents and visitors to their destination safely, we need to ensure that the nightmares of AVs that we are seeing all we're seeing in other places don't become our stories as well.
Don't allow the replacement of safe professional drivers with unaccountable and incomplete AVs. Thank you. Thank you, sir.
Speaker 8
Alright. My name is Mustafa El Amin. I'm also a I'm representing the I'm with the California Gig Workers Union. I am also from Fresno, California. I've been a rideshare driver for going on ten years now.
So as a member of the California Gig Workers Union, as a rideshare driver, you know, I take pride in, transporting my passengers in a safe manner, providing an excellent service that I can provide for them as well. And AVs, know, they're not been in my released in our area as of yet, but it's coming. And I already know that, you know, I'm not against innovation, but, you know, right now, innovation is kind of outpacing the legislation. So there has to be a framework for things to be rolled out in a safe manner. As a rideshare driver myself, there's a certain criteria that I have to abide by.
I have to meet the legal framework that that's that's put on myself, and I have consequences if I violate any of these types of rules. So just we just wanna make sure that AVs, as they get rolled out, they have the same considerations in mind, you know, especially if it comes to the we're operating in the same space that I operate in as well. So we wanna make sure that, you know, the supporting bills that are gonna create a legal framework so that AVs have the same accountability as the human drivers.
Speaker 21
Chair members, Alicia Priego here on behalf of the Chamber of Progress. Thank you for this important hearing. AVs are a part of a broader mobility ecosystem that augments and expands good jobs across our economy. With smart, safety first policy and intentional workforce development, California can capture these benefits for working families across the state and continue to lead with policies that support forward thinking innovation. This shift won't happen overnight, and rollout is expected to unfold over a decade or so, giving time for this transition.
AVs create indirect employment across infrastructure and construction, energy and utilities, local services, and expanded logistics and retail activity. I did want to touch on one issue that hasn't been covered today is that AVs can help transform paratransit, offering flexible curb to curb service that could help as many as 4,400,000 people with disabilities access jobs nationwide. The Public Policy Institute of California has emphasized that reducing transportation barriers, expanding job training, and strengthening employer supports are central to improving labor market access for Californians with disabilities, and these are goals that AVs can help advance. Thank you for your time, and I appreciate the ability to comment.
Speaker 22
Next speaker. Thank you, Vice Chair and members. My name is Ashanti. I'm the Director of State Policy here at the Silicon Valley Leadership Group. SVLG represents Silicon Valley's most innovative companies, including leaders in autonomous vehicles.
We're here today to urge the committee to recognize the distinction between fully autonomous vehicles and driver assist systems. This distinction matters. By the state's own determination, vehicles with level two driver assist technology are not classified as autonomous vehicles, and conflating them with level four, fully driverless vehicles misrepresents the technology, the safety record, and the regulatory framework. The safety data is clear, published from Waymo. Published data from Waymo shows that across San Francisco, Los Angeles, Phoenix, and Austin, and across all crash types, fully autonomous vehicles are delivering 80 to 100% less reductions, caused by distraction, impairment, and speeding, leading causes which are the leading causes of death for roughly forty thousand traffic deaths, Americans suffer each year.
These results are consistent across all four cities and all severity levels. This is not a localized advantage. It is something that is a systemic gain. We believe that this represents the most significant advancement in road safety in recent memory. The regulatory framework matches that promise.
California is building the most rigorous AV oversight in the country with the DMV finalizing autonomous truck regulations that go above and beyond what the legislature had previously required. A time where the legislature is up, please. Rightly focused on road safety, Silicon Valley is delivering real, measurable solutions. We ask the committee to let the data guide the policy and protect California's role as a global leader in autonomous vehicles.
Speaker 13
Alright. Thank you. We are rapidly running out of time. So, if we can get your comments in, I appreciate it and keep them succinct. Thank you.
Speaker 23
Thank you. Elmer Lazarda here on behalf of the California Federation of Labor Unions. Just want to state, you know, that there are over half a million workers who drive for living in California, especially in delivery, freight, passenger service, public transit, etcetera. And commercial drivers have the expertise and training to navigate the complexities of California's roads because they are licensed to do so. And yet, companies are looking to increase profits with AVs at the expense of those by attempting to eliminate jobs entirely.
Driverless vehicle deployment has been expanding in very visible ways, as has been mentioned throughout today. And we want to ensure that as AV technology is expanding into more industries and more heavy duty uses, people are aware of the threat of the job loss and economic displacement that is becoming more pressing, especially industries like heavy duty delivery, agriculture and construction. We at the labor fed want to ensure that we are updating our laws to meet the challenges of our twenty first century, especially as they relate to workers' technology rights and how those technology and technological advancements are changing the scope of work. And AVs are no different. So, we just want to ensure that workers remain at the forefront of these discussions and that we relate that technological advancements are not new.
We've heard about the advancements of so called innovation. But it is the workers who are going to be bearing the brunt of these changes, and we want to make sure that if we care about protecting workers and their dignity, that they are at the forefront of these discussions.
Speaker 24
Good afternoon. Matt Hetich with the Transport Workers Union. Thank you for an opportunity to discuss AV policy in California. The Transport Workers Union represents bus operators and dispatchers and other transit safety professionals at SF Muni. While the Transport Workers Union is not opposed to driver technology that aid a driver in the performance of their jobs, we welcome blind spot monitoring, lane departure, and auto braking technologies, we are opposed to technologies that replace a human operator, a well qualified human operator.
Bus and transit operators provide critical safety component in our city buses in San Francisco. These are good middle class and union jobs, and our members wish to continue in these roles as safety professionals for the future. We thank you for your time.
Speaker 13
And thank you very much. That exhausts our speaker role. So I've been asked to go and put the committee in recess until the chair returns so he could potentially whoops, is he here? There he is. And here's the chair to ask you questions.
Thank you very much for your guys' attention and your patience today in testifying on this. And with that, Chair Cortezi will be finishing up. We've already gone through public testimony.
Speaker 1
Okay. I'm at a disadvantage of the legislature's own making here, not having heard your presentations. I apologize for that. I'll go over them afterwards digitally because they are important to me. I do I want to ask you kind of a question that is really general and it's you could know is a perfectly fine answer.
But you heard some of my questions if not all of them at least by sort of the osmosis of being in this room to the previous panels in terms of things like response times and those kind of regulatory issues. Is there anything there that you would comment on either from the DMV side or the CPUC side? I guess what I'm saying is I would certainly welcome your thoughts on any of those questions that I brought up in terms of the need for uniform standards, regulation of backroom operations, response time issues, and whether there should be affirmative. I guess I was implying maybe there should be some sort of affirmative dispatch communications. Has any of that been, you know, come to your attention in your own agencies?
I can certainly
Speaker 14
start, Mr. Chair, and then Miguel can provide some of the details. But we are aware of the incidents that have occurred and Miguel can go into the actions that we have taken. I will say that I'd like to point out that we're not adverse to pulling the levers that are available to us as far as enforcement. For example, you know, we took action against Cruise when it was appropriate to do so.
We took action against other companies as well. So it's not something that we take lightly. We take our responsibility very seriously to ensure that the motoring public is safe and we feel like the regulations provide that vehicle for us to do so. That being said, these incidents that have occurred are very much on an operational level and we also have looked into those incidents and worked with the companies, worked with the first responders, worked with the people who are affected by them so that we get a clear understanding about what had happened. And then that helps us helps inform us moving forward.
So,
Speaker 15
like to add that the new regulations that we're going to be we've developed and we're looking to actually adopt have additional enforcement tools. So currently, right now, we have the enforcement tool of suspension or revocation of a permit to the operating authority. We're actually looking for additional enforcement tools that would allow us to actually potentially restrict, so give us a little bit more ability to target potentially certain areas of an operational design domain, geographic location, and give us, again, more tools to enforce incidents occurring in a variety of ways.
Speaker 16
Thank you for the question. I think the only other thing that I would add is and you had missed the earlier presentation, but appreciate the question and the opportunity to clarify the Commission's regulatory role I over autonomous vehicle passenger services would say that a lot of the issues and topics that came up in the discussion earlier today are issues that we have heard from the public as well. And so I thank you again for the opportunity to kind of clarify our role and the DMV's role over those types of topics.
Speaker 1
Just a couple more specific questions, and I appreciate the two. Like I said, I'll go over your prior comments. My own observation, and I alluded to it earlier, is that since the days of Henry Ford, there was this combination, and it probably didn't go exactly back to that date, but generally speaking, going back to the time of the mass production of the automobile, pre autonomous vehicles, obviously, there was always been this blend, as far as I know, of state agency regulatory control, especially over licensing and registration, the ability to revoke licensing and registration, which still exists today. We've had DMV here and talked about those things recently. And yet there's still, despite the state agencies occupying so much of the field, there's still this local regulatory process that basically seems more operational in many ways, right?
What do you do? Can you make a right turn from somebody was talking about seeing an AV making a left turn or right turn from the center lane. Typically, those may or may not be state motor vehicle code issues, but typically the enforcement and the local ordinance making around whether you do that in the airport approach or whether you do that on a particular frontage road or street, in my experience, is in a stack of local ordinances that have been adopted by many cities within the framework that the DMV allows and perhaps
Speaker 1
CPUC on things like taxi service and things like that. Are your agencies themselves opposed to some level of complementary local ordinance making, rule making, regulatory process on just day to day congestion and driving issues. And I get it you may not be able to speak on behalf of the entire, I'm not asking you to speak on behalf of the executive branch, being an executive agency here. I'm really just asking is there, would there be a fundamental reason with autonomous vehicles you know, to not have that statelocal blend of control? Is there something I'm sort of missing there?
Speaker 14
You're right. It's a question that's difficult for us here to answer on behalf of the executive branch. And from the standpoint of what we are responsible for is ensuring that those vehicles are safe to be operated on California's roadways. All California roadways. So we have to ensure that what we put down in a regulatory fashion does exactly that.
And so we look at things such as obeying all of the rules of the road and at a state level. So if there are some items that are at a at a local level that also need to be conformed to, you know, that would be an expectation that we would also have because it encompasses the safe operation of those vehicles. No different than us licensing Miguel, myself, yourself Right. And the expectation that we would have that you would obey the rules of the road regardless of where you are in the state. Right.
We
Speaker 1
were having a little sidebar, a couple of us over here just before the hearing started, that this is one of the few areas of machine learning, AI related stuff, if not completely AI driven technology that's coming along, emerging, and already on our streets. It's also an area that we've regulated before. So much of what we encounter, again speaking as one legislator these days, falls into the category of never seen it before and we really have to start with a blank slate. Maybe like everybody did in Henry Ford's day, how are we going to deal with this? But we have this experience with motor vehicles in terms of the regulatory side of it where we've had that blend.
It seems to have worked well. It seems, even with us very deficient, less than 92% safer human drivers. When I pull over on Highway 5 into Stockton, I know there are different rules, local rules, than when I pull off in Tracy or eventually make it all the way to San Jose. I mean, my brain is capable of computing at least that far. I know the basic rules, that the DMV is promulgated, and I don't have to worry too much about CPUC.
But I also know what to look for as soon as I come into a local town or a city, a rural road, a scenic highway. I know what to look for. It seems to me autonomous vehicles can adapt local nuances as well. But if you disagree with that, and that's more of a technical question, not a philosophical question about the administration of the executive branch or your agency, that there's a place that sounds like what you were saying is that there may be a place for local rule making of some kind to complement the larger process that you're in charge of.
Speaker 14
Yeah. Mr. Chair, don't disagree with you at all. I'm certainly not an AI expert, but I would anticipate that through machine learning and different AI technologies that that learned behavior, that learned methodology that you and I go through would be transferable in the machine world. Not sure exactly when that would happen, but I can envision that that would be the case.
Speaker 1
All right. We'd love to certainly have a cooperative relationship with your agencies. I think I can speak on behalf of the committee, at least, here in terms of flushing out those kinds of issues going forward much easier and much quicker to work collaboratively than and that was the nature of my question. Is there potential for collaboration there with the DMV, for example, on trying to get there as opposed to the legislature itself post some kind of a tragedy or, you know, based on some of the fact patterns that we have, just putting legislation out there that may or may not align with your thinking as to what should happen in terms of the local side of things. Know, having come from local government, I'm very confident in cities and counties' ability to deal with the nuance of their own congestion management, where taxi stands should be, where rideshare should show up or not show up, seems to be very difficult for
Speaker 1
to, either you or the legislature, rule make or to legislate from this level, from a singular capital in Sacramento. Any further comments on that?
Speaker 14
I would say we would welcome any discussions that you would want to have.
Speaker 1
All right. That's appreciated. Very much appreciated. Have there been the last thing is, I mentioned it in my first broad question, but have you had discussions specifically about maybe what should happen in terms of uniform dispatch and response standards when it comes to first responders? We had a lot of first responder testimony today.
I'm just wondering if that's come up in either of your agencies as something to look at.
Speaker 14
We have. We have had discussions with several first responder organizations. Miguel, if you want to go into some of the details with the I ones we
Speaker 15
That's fine. No, we've hosted, in the development of our new regulations, we've facilitated two statewide workshops with first responders, first in Northern California and then some in Southern California, listening to some of those exact concerns. In addition, in the development of our regulations, you were mentioning best practices, what kind of standards. And so in addition to hearing input from all stakeholders, we took a look at what industry best practices are available for those kinds of interactions, particularly when there's no driver in the vehicle, and how to interact with remote assistance. So again, I think our approach was to not only hear from the first responders on the ground who are dealing with those in the cities, but also to the stakeholders who are developing the technology and seeing the best practices across the industry as well.
Speaker 1
And any plan yet for how to deal with the actual individual, this comes back to the local side of things, the actual individual dispatch centers. Meaning, obviously first responders are dispatched now from these sort of centralized, especially in the urban areas, but centralized dispatch centers that are getting police and fire and ambulance out on a 09:11 call. Any thoughts yet or developed yet in terms of how you might or how we might interact with them. Again, does it come into the area of they know what they're doing, they should be empowered to work within a framework that we set up, or we'll tell them exactly what we need and put that in our regulations? Or none of the above?
Speaker 15
I think one of the things is we're hoping to get some more data on that. And as these AVs scale in cities like San Francisco and Los Angeles, trying to see the impact, we're going to continue to learn and adopt regulations that are going to continue to grow with the industry and also the needs of the first responders. Okay. Last
Speaker 1
question. The different nomenclature that I was questioning before, I was hoping by the end of this committee hearing to understand it all better. I think I accomplished that much. Has there been any discussion about just getting the nomenclature on the CMV? I understand about the Tesla, the modification to FSD and making sure it's labeled as unsupervised or rather supervised.
But it seems like at some point this should all be consolidated into basically a couple of definitions, a couple of glossary terms in terms of what we're dealing with out there. Because there's a lot of redundancy in the four or five terms that we heard. Is that something that you're working on?
Speaker 14
It certainly is something that is an area that needs to be looked at. We have to be also mindful about coming up with definitions that may not comport with other jurisdictions throughout the country. And so it's the definitions that exist came from very much an engineering background because of the way that the the systems were being developed. Not necessarily the most consumable terms for the general public. And as such, as technology develops, I think there is a need to have a standardized nomenclature that is easily understandable, consumable, and one that most people would understand and it makes sense when you hear it.
Speaker 1
Yeah, CPUC's dealt with a lot of that, a lot of nomenclature around other areas of utility regulation and so forth. Is that something that would find a home at CPUC? Type A, type B, type C, and that's it? That's what we have out there?
Speaker 16
Yeah, appreciate the question. I think when it comes to classifying autonomous vehicles specifically in California, we look to the existing law and the DMV's regulations, I will say, I think one of the terms that was used quite a lot today was robotaxi, and I think that is colloquially used to refer to passenger service and AVs, though it's not a regulatory term. And so that is something that when we get inquiries from the public, from the press, from others, we try to clarify and use the standard regulatory terms. But agree with Bernard, they're not always the most clearly understandable for the public.
Speaker 1
Yeah. Just an example of that anecdotally is during the interim recess, kind of in trying to do some homework before we started up this new year, I paid a lot of visits to manufacturers. And without naming names, one of the things that I was told during one of those visits was we, manufacturer X, are going to see this point to a vehicle. We'll be using this as a robo taxi in conjunction with another firm that's already in the passenger service. So we're entering not a merger but a contract to deploy what we manufacture to a company that's in the service delivery side.
They refer to it as a robotaxi even though the car already has a name, a model, and everything else. That can be confusing from a consumer protection standpoint, especially given if a company's taking what you guys would call level two in referring to that. And I know, again, that that's been a problem, as we just talked about. How much of a problem is, sorry about this as being the last, last question, but how much of a problem with all of this is, hey, it's important to honor manufacturers terminology, but it's even more important to utilize consumer protection against marketing something that isn't really what it is. And I mean, if you can adopt those rules, there's plenty of attorneys out there who will enforce them under consumer protection laws.
But we haven't seen much of that. And I guess that's what I'm asking for, asking about. Does that come from our side? Or does that come from your side? Legislative or agency side?
Speaker 14
Right. Mr. Chair, you are aware of the actions that we've taken against Tesla? Yep. Because of the nomenclature that was being used.
You know, that is an example of one where, you know, we felt that, you know, the words matter and the way that someone labels their product may not be indicative of what their product is capable of doing. So again, it feeds into what you had just mentioned. The need to have some sort of nomenclature that would be consumable by everyone.
Speaker 1
Editorial comment before we close. Often it's been the case in California when it comes to motor vehicle regulation of any kind, including air quality issues, that we can't wait for a national uniform policy nomenclature. We have to be the ones to establish that, basically by our own actions. So hopefully that kind of thinking prevails here and we're not both on the agency side and the legislative side. Thank you very much for being here.
Thank you to the folks who stuck around. I understand public comment's already done. Again, I apologize for the delay that I fell into running over to vote on four audits in the other building. But all that's done now, and we appreciate your time. And I hope everybody enjoys the rest of day.
I want to thank our committee staff for organizing all of this. I think it was useful and will help people very much here down the road on the policy side.
Ask anything about this video — I analyse its transcript to answer.