Committee on Constitutional Affairs Ordinary meeting ICM
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
AFCO European Parliament
Follow its sittingsRoberta Metsola
It's truly great to be here to open these AI parliamentary days. Let me also welcome all of you who are joining us from the national parliaments across Europe. There is an old line that everyone in this room has probably heard more times than they care to remember, and this is that the US innovates, China replicates, and Europe regulates. Now it was a good soundbite, but it hasn't aged well. For one thing, China is now competing at the very frontier of AI, but the picture in Europe has fundamentally changed as well.
And I want to use these few minutes, if you allow me, mister chair, to perhaps push back on a few things that we still hear being said about Europe and AI that I think are simply not true anymore. First of all, you mentioned this in passing, Sven, is that, I would like to push back on the idea that this parliament loves regulation for regulation's sake. We wrote the AI act not out of habit, not to make life harder for anyone, but because we are clear-eyed about what can go wrong. Disinformation that distorts elections, deep fakes of leaders almost impossible to tell from the real thing, authoritarian governments using AI to monitor and control their own people. So people have a right to privacy, a right to know when a decision about their life has been made by an algorithm.
And these are real risks, and they deserve real answers. And I wanted to start with this because when we ran for elections in 2024, and I'm sitting between Gabby and Sven, we would go and wherever we would, we would have to almost justify why we had to pass this act. And I think when we look now around the European Union, but also around the world with so many pieces of legislation that actually stemmed from RAI Act, I think it would be good to take pause and say that this is when we really courageously made the decision, and I think it was the right one. Just to give you an example, in the past year alone, individual United States states have passed 18 new AI laws. And I'm saying this because everyone is grappling with the same questions that I just laid out right now.
Now, as an aside and to go back on what I just said, for rules to actually work in the real world, they have to be, first of all, smart, and they have to be proportionate. And at the same time, they cannot be so rigid that they slow down the very innovation that we are trying to foster in our continent. And the members of understand this, which is why just a few weeks ago, we voted to simplify the AI rules. And that shows that we listen, we adapt, and we are not afraid to change course when it makes sense. The second thing I would like to push back on is the idea that Europe is just a spectator in this AI revolution, and that in a way we are sitting on the sidelines while others are building the future.
In fact, if we look at the situation in each of your countries, the picture on the ground tells a very different story. Recently, Jan LeCain, one of the people who literally invented modern AI, someone who would have been welcomed with open arms anywhere in the world, chose Europe and he chose France to launch his new venture, raising over a billion euros and the largest seed round this continent has seen. Meanwhile, we have Mistral, who is building a massive new data AI center in Sweden. And these are not isolated decisions. They are part of a broader shift, and they are also part of a growing recognition that, first of all, Europe has the talent.
It has the research culture and the serious long-term thinking that this technology needs. And we are backing that in very real ways. We have mobilized €200000000000 for Europe's AI industry. We are building gigafactories and data centers, and we are pushing hard on permitting because, also, let's face it, navigating 27 different national systems with their own rules and their own timelines is slowing us down in ways we can no longer afford. On energy, of course, against a very difficult background right now, we are investing and modernizing our grids, but the pace is not matching the ambition.
We know AI needs power. It needs lots of it. And if our data centers cannot get the grid connections that they need, then everything else will hit a brick wall. And then there is a human side because new jobs will come in data, in model development, in human-AI coordination. But managing that transition will not happen by itself.
Workers, whose roles are changing, and we are facing this very challenge in our own institution, need answers, and they need real investment in reskilling. Children need to be prepared for an economy that will look very different in five, ten, imagine 20 years. And I think these are questions for all of us as legislators wherever we sit, in this parliament, in a national or regional assembly, because making sure that this transition works for everyone and that everyone is part of it and that no one is left behind is fundamentally at the heart of everything we do, which brings me to my final point. Ultimately, the decisions that will shape Europe's place in the AI world are not only made here in Brussels. They are made in capital.
They are made in regional governments and national parliaments, and we need all of you. We need all member states moving as much as possible in the same direction on investment, on infrastructure, on job security, on cutting red tape, and we need parliaments to drive that change and hold governments to account. So if I can, I started with a cliche, but I'll finish with another one? So if we want Europe to be the next Silicon Valley for AI, I think we all have to pull together, we have to do it now. So I'm really grateful that this week has been organized.
I'm really grateful to all of you for taking the time and spending it with us, as legislators in this house, because I think together, this continent can do really great things. I'm glad we can do it. Thank you.
Sven Simon
Thank you very much, Madam President. And I think now since the commission she's
Sven Simon
Three minutes? But then I think we give the floor first to our keynote speaker. It is a privilege to welcome you today, mister Amandeep Singh Jil. He's the United Nations Undersecretary-General and Special Envoy for Digital and Emerging Technologies. Your leadership undersecretary general at the United Nations places you at the center of these reflections, and we very much look forward to your insights.
Amandeep Singh Gill
Thank you, mister chairman, madam president, honorable members of the European Parliament, honorable members of national parliaments, excellencies, ladies and gentlemen, thank you for this opportunity to exchange with you today. And, madam President, thank you for setting the scene so clearly and boldly. Artificial intelligence is no longer an issue about the future. It is about our present. It is transforming our economies, our societies, our public services, and soon, our political systems.
So no country, no matter how advanced, can fully manage this transformation on its own and without investments in some critical areas. And yet, this has to be done at a time which is not easy, when multilateralism is under strain, public institutions have been underinvested in for decades, trust is uneven. And citizens, especially younger citizens, are uncertain about the future, and AI is playing a part in that uncertainty, in that anxiety about the future. So existing frameworks must evolve. We must also strengthen and preserve what works.
No digital or emerging technology challenge can be handled by working in the digital space alone. Our analog institutions, parliaments, court systems, schools, they all have to be strengthened alongside. And here, I submit to you that international cooperation would be essential. I'm going to address three broad themes in the next few minutes. The first one is on what is it specifically about AI that calls for a coordinated international response.
First, AI is truly global, truly transnational in its evolution, in its impact. Let's not have any illusions anymore. It's a powerful general purpose technology, but perhaps very different from general purpose technologies we've known in the past: the railways, the steam engine, the computers, and so on. It evolves rapidly and across borders by design. Models are trained across jurisdictions.
Data flows globally, and the impact of AI models is felt everywhere. A system developed in one region, you know, take the recent example of the step change in the capabilities of the Claude model, impacts financial markets, stock valuations, labor markets, and decisions about investments everywhere. Second, AI brings both opportunities and risks. So it can boost productivity, accelerate economic growth, and transform sectors such as health and education. At the same time, it introduces new risks.
The president spoke about that, to security, to the integrity of public institutions, to democratic processes. I recently had the honor to engage in an exchange of views with the special committee on the EU Democracy Shield here at the parliament, to discuss the risks of the unholy trinity of misinformation, disinformation, deep fakes and reality distortions, and hate speech. The third aspect to bear in mind about AI is that it concentrates capabilities and power in very few hands. And this is not only a geopolitical question. It's a question within national jurisdictions of a shift in balance of power between public institutions that are voted, that are elected, and the power that's held in the private sector.
So, of course, it drives shared prosperity. It can also deepen inequality and power imbalances for generations. We've already started hearing about a K-shaped economy, a K-shaped society. So there's a real risk of an AI divide, and this is not a North-South, global North, global South issue. It's a global issue, period.
So if left ungoverned, AI growth will not only deepen inequality, but it'll also amplify, it'll accelerate it. So these features, these three features of AI demand a coherent, cooperative response. And in that regard, the engagement of European institutions is particularly important, particularly valuable. So let me come to the second part on Europe's valuable role and the need to build shared trails. Europe, of course, has been at the forefront of shaping a values-based digital agenda.
It brings critical experience in protecting fundamental rights, ensuring accountability, and building public trust in digital systems. At the same time, Europe, like others, is navigating how best to balance AI regulation and innovation. Regulation that requires constant adjustments, so finding the right balance. So this means smart regulation, agile regulation, and distributed regulation. Whatever approach we take, and we see at the UN different approaches across different jurisdictions, some law based, some based on executive decisions, some based on a degree of self-regulation, whatever approach is taken.
I think what's fundamental, what's important is the need to preserve core principles, respect for human dignity, the rule of law, and the general public interest. So I firmly believe, and my boss, Antonio Guterres, firmly believes that Europe's leadership matters, not only as a model to learn from, but also as a key contribution to a broader global conversation on this values-based AI governance. Otherwise, even the most advanced approaches, if they are only limited to one geography, will have limited impact. So when I say shared rails, common understanding, what do I mean by that? Common approaches to AI safety and risk management for the most vulnerable.
The risk management approach in the EUAIA Act is a fundamental principle that's kind of transcending borders. Madam President, you referred to that. Second, interoperable standards and technical benchmarks. Third, alignment on transparency and accountability practices. Fourth, and this is important, a mechanism to share resources, knowledge, and best practices.
Conveyor belts that work across parliaments or executives to share these best practices, these lessons learned. So these are necessary conditions to build trust, boost innovation, and share the benefits of the AI transformation. And for the final part of my remarks, let me turn to the United Nations, the platform for global digital cooperation. So I'm often asked, you know, how is the architecture of digital cooperation going to evolve? Will it rely on treaties, on formal agreements, or more agile multistakeholder coalitions?
The answer is both. And from the outset, my guidance from the secretary general has been to find the right articulation between hard norms and soft norms, because treaty making takes a long time and is very difficult in the current times. So we have to be smart about how we bring in the global element into this conversation. At the foundation of this is the fact that the UN brings all countries, big or small, to the table. And there is a key.
There is a heritage of international law, the UN Charter, the Human Rights Treaties, international humanitarian law that's already connected and available there. And so over the past few years, we've been going about this with some strategic perspective, of this articulation of soft and hard norms and these principles of inclusion and linkage with international law. So let me give you the most concrete example, and this was the adoption of the Global Digital Compact in September 2024. In the final negotiations, the European Union played a decisive role in the adoption of the GDC. This has the common vision for an open, safe, and inclusive digital future.
And for the first time, for such an agreement at the UN, it has decisions on the governance of AI. And those decisions have been worked into practical modalities over 2025. And what are these modalities? First of all, the independent international scientific panel on AI of 40 experts, of which a dozen are from the European Union, co-chaired by Yoshua Benjiu, the Canadian Turing Award winner, and Marie Ressa, the Nobel Prize winner from The Philippines. And this panel will bring regular scientific assessments and evidence to the policymakers, including parliamentarians.
What's the direction of the technology? What are its impacts? What are the opportunities and risks that are associated with it? So we see what's coming around the corner, and we can adjust our policies accordingly. Parallel to this, the twin sister of the science panel is the global dialogue on AI governance, where all 193 member states have a seat at the table.
This is democracy at work at the global level. If AI is going to impact us all, all of us must have a say in how the rules of the road are shaped. So this is a structured process which brings policymakers and other stakeholders together to share lessons, learn from each other, build common understandings, build interoperability across different jurisdictions with the background of that independent evidence-based insights from the science panel. Then we have a third component, which is capacity. So some call it the AI divide, which is very real.
I think capacity, the capability to work with AI, is crucial. Otherwise, governance will be symbolic. So you cannot govern what you don't know, and what you do not develop yourself, you cannot own. So only by having some understanding of how this technology comes together, what are the different pieces that are required, the critical enablers around data, infrastructure, talent, etcetera, can you actually understand it and then govern it well? So it's a fact that today, the AI capabilities are highly concentrated, data center capacity is highly concentrated in US and China.
All of Africa, less than 1% of the data center capacity. All of Africa last year, less than 1,000 GPUs. A single university in China, which is not a leading AI university, Fudan University, has 1,200 GPUs. And I'm not even talking of companies like Meta, which have 600000 GPUs. So this divide is real.
And at the same time, we have connectivity gaps. We have energy deficits around the world, and billions of people remain without reliable access to the internet. So cannot have a world, this K-shaped world. If you can't have k-shaped societies, economies inside, we can't have a k-shaped world outside. So we need to work hard, and the secretary general has presented proposals from member states to address the AI divide.
So international cooperation is a must across these areas of science, policy, and capacity. Now I wanna conclude by just recalling why this matters for Europe. Engagement in global cooperation, multilateral efforts, is a strategic opportunity for the European Union, also to connect with the global South. It allows the European Union to expand the reach of its values and principles, to reduce global fragmentation, strengthen resilience, and to build trusted partnerships across regions. And most importantly, from
Amandeep Singh Gill
Perspective, it helps us to ensure that global frameworks are shaped by shared values and the global public good. Thank you.
Sven Simon
Thank you very much, Under Secretary General. I will now give the floor to Miss Hena Virkonen, a former member of the European Parliament and today, the executive vice president of the European Commission for technological sovereignty, security, and democracy. The floor is yours, Hena.
Henna Virkkunen
Thank you very much, honorable chair, honorable president of the parliament, under-secretary general, honorable members of parliament. Thank you very much for inviting me to speak in this inter-parliamentary committee meeting on artificial intelligence, a key topic for our future as we know. And as the chair already mentioned, having served as a member of this house as well as a member of the Finnish parliament, I especially appreciate this kind of interparliamentary exchange. When I became executive vice president in charge of technology, I started with a vision of making Europe an AI continent. We know that for us in Europe, it's very important that we are creating trustworthy technologies, and we already had the AI Act in place, and it's creating trust and excellence as we have seen.
But Europe still needed a stronger boost for innovation and competitiveness. And when we launched the AI Continent Action Plan, that was a very clear topic for us. With our team in the AI office, we built AI Continent around five pillars to give European AI the support it needs: infrastructure, data, skills, adoption, and simplification of rules. The AI continent is not only about building AI in Europe, but it's about building it in the European way, so centered around our values, high-quality research, innovation, and industrial know-how. And in the past year, we have already seen key parts of the action plan also delivering results.
We have boosted our infrastructure with 19 AI factories. So together with our member states, we have decided to build 19 AI factories, because we see that we really need that kind of world-leading supercomputers in Europe to start building their innovative AI models, because this has been a clear obstacle for our AI development. We know that we have very big potential in our startups. We have currently 8,000 startups who are developing and training AI. But one bottleneck for them is that they don't have access to computing capacity.
And that's why these investments, what we are making with our member states, are crucial for our start ups really to create AI models, but also our researchers to have access to computing capacity, what is needed to create AI. And we have also set out the AI gigafactories as a way for organizations and researchers to access the computing and data capacity they need to train very complex AI models at an unprecedented scale. And we will very soon also open the official call for these AI gigafactories. But currently, we know that there is now much more demand for computing capacity than what we can provide, and that's why these investments are very important. But at the same time, it's important to acknowledge that infrastructure alone is not enough.
So we also need AI adoption. And this is about enabling our strong traditional industries, such as our pharma industry, automotive, and robotics, to shape the AI revolution, generating value for our economy and society. And, of course, the public sector as well, there we have huge potential also to use AI. And for this, we published the apply AI strategy, encouraging organizations to think AI first and integrate AI technologies into their ways of working. Our support for AI adoption is already materializing with €1,000,000,000 of funding calls mobilized to support AI uptake in European strategic sectors.
And in addition, we are rolling out targeted sectoral initiatives now. For example, in the health care sector, we are setting up a European network of AI-powered advanced screening centers that will support prevention, early detection, and diagnosis of cancer and cardiovascular diseases. And this call is already open, and we hope that we will have participants from each member state. And on top of this, the Apply AI strategy is also institutionally a strong governance mechanism. So it supports monitoring of the impact and uptake of AI in specific sectors through the AI Observatory.
And very importantly, for AI continent, we also need an ecosystem of trust across the EU. Trust in society is a precondition for AI innovation as we know. So our AI Act is delivering on this through a single European legislative framework, offering legal certainty across the single market. Because like it was mentioned already and underlined here earlier, we have great, of course, opportunities with AI, but of course, we have to be aware of risks as well. And that's why we have this risk-based approach with our AI Act.
It's a first-of-its-kind law setting a global benchmark for AI governance. It follows this risk-based approach, introducing rules only when we see that there is risk. And it creates trust, while leaving room for innovation. And we are very strongly committed to the effective implementation of this AI Act. This AI Act applies in stages.
So the prohibitions and rules for large models, they already apply now. And other core provisions like labeling of deepfakes are set to apply now in August. So step by step, this AI Act is coming into force now in the coming months and years. The AI Act amendments we have proposed in our digital omnibus address implementation challenges and simplify also our compliance, and they aim to make sure that we land the AI Act well without lowering the level of protection for fundamental rights. So in the same time, when we want to really boost innovations and investments, we also see that kind of car trails are needed, and that's why we have our AI Act in place.
And we are also implementing it in a very innovation friendly manner. A key factor for our European approach to AI is strong coordination, oversight and expertise, and this is where the AI Office comes in. The AI Office was established in 2024 based on the AI Act. And today, it supports the EU's AI approach with 140 members of staff, including technology specialists, ethics specialists, and also AI lawyers. And I have, of course, read with great interest to AFCO a draft report on the institutional aspects of AI, including the vision to establish a European Parliament's own AI Observatory.
I agree with you that this house, as well as the national parliaments, have a very crucial role to play in common work towards AI continents. And it is important that we also have very strong inter-institutional coordination together. At the same time, on a very fast-evolving technology like AI, it is important that we have a European central point of expertise as well, so that those who deal with AI governance and innovation policy always see the broader picture. I already mentioned earlier the role of the AI observatory in this regard. And this is important also from the international point of view.
The AI office is steering the EU's international activities in the field of AI and serves as a point of contact for international partners, like for the United Nations, including for the recently ratified Council of Europe's Framework, also convention on AI. The AI Office has already become a global reference point for AI policy, which we can be very proud of. So to conclude, a lot of work has been done over the last years to really shape European AI policy, and we are now in a phase where ideas are mattering into very concrete actions. I appreciate occasions like today where we can jointly also reflect on our next steps to make Europe an AI continent together. Thank you very much.
Sven Simon
Thank you very much. So now we have had inspiring keynote speeches and have set the scene, so to say. We will now continue with a tradition which we established in the Constitutional Affairs Committee recently, I would say, namely that we connect the inter parliamentary committee meeting always with a report so that you, as members from national parliaments, have really an influence on our report and the rapporteur will listen carefully. I think we have done that once, and this is the second time today in the history of the European Parliament in relation with national parliaments. And I think it makes sense, so that I'm very happy that many colleagues from national parliaments are present here.
Thank you for being here.
Thank you.
And you
have to leave us there. Yes. And I.
Sven Simon
Yeah, then we can continue with the consideration of the draft report on our own initiative report with the title Institutional Aspects of Artificial Intelligence in the Context of European Integration. And here we have a rapporteur, our colleague Emmanuel Kefalong Giannis from EPP. Thank you for taking responsibility here with this report. I know that it was very important for you to be the rapporteur. And now I give you the floor to introduce to us what you have done so far.
Emmanouil Kefalogiannis
Thank you, chairman. Ladies and gentlemen, and colleagues, artificial intelligence doesn't concern only technology. It concerns the way in which we govern, we legislate, and we get informed. And finally, the way in which the democracy itself works in the European Union. Our report is based on four basic axes.
First of all, artificial intelligence should remain aligned with the fundamental values of the European Union, with Article 2 of the treaty as a compass, the article that defines the fundamental values of the European Union. Artificial intelligence should serve the human being and not replace the human being. It should support and strengthen democracy. And because in democracy, the decisions would be based on the popular vote and will, and not be replaced by algorithms. Artificial intelligence has to work with transparency with respect to the rule of law and to fundamental rights.
Second, we propose to increase parliamentary control and accountability. We propose to create a European Observatory of Artificial Intelligence in the European Parliament, not in order to regulate or oversee, but to support the parliamentary work, the institutional memory, and the democratic control without any economic cost for the European taxpayer. Third, concerning artificial intelligence, we need a wider global framework of rules. We need an international convention for artificial intelligence as we had in the past at the end of the decade of the '60s with the National Convention for Nuclear Arms, and always in the framework of the United Nations. Europe cannot be an observer.
In this new reality, Europe has to be the star. Finally, in the European Union, we need now already a serious institutional discussion concerning whether the whole institution, the European institutions, are protected sufficiently for artificial intelligence. The governance of artificial intelligence may size the efficiency of the treaties of the European Union. Ladies and gentlemen, colleagues, in the era of artificial intelligence, democracy and elections are not only threatened by the lie and populism, they are also threatened by targeted, algorithmically supported misinformation, for example. Deepfakes, they're not just a side phenomenon.
It's a weapon trying to manipulate public opinion, and it tries to destabilize the democratic procedure. That's why we need clear rules, operational readiness, and technological tools in order to be able to spot these problems like watermarking. E, be able to have a traceability to have a stamp of the contents produced by artificial intelligence, as defined in the AI Act. This is an important first step of progress. Ladies and gentlemen, colleagues, in the era of artificial intelligence, democracy cannot remain without being strengthened.
Europe has to move ahead with rules, with institutions, with responsibility, with transparency. Thank you very much for your attention.
Sven Simon
So next speaker will be the rapporteur for the opinion of the IMCO committee. It's our committee for the common market, Mister George Pipereira, please.
Gheorghe Piperea
Thank you, chair. I'm gonna speak Romanian for better understanding.
Dear colleagues, thank you, first of all, to mister Cefaloganis, the Afco rapporteur, and to his entire team for the very good collaboration we've had today. We are assisting to a very important debate on AI. I support this report's objective, namely to consolidate our continent's capacity in the field of AI. From an institutional point of view, it is essential for the European Parliament to retain its democratic legitimacy. We can see how algorithms can influence political choice.
We need to be very careful about this field. Updating the EU electoral framework would be a good idea to reflect all of these new realities. A parliamentary observatory for AI would be useful, but only if it remains a support mechanism, not a regulating mechanism. We need to remain mindful of fundamental rights and values that need to be protected against AI. We need to focus on who makes decisions, how it is explained, how you can fight against it, how responsibility is ascribed, etcetera.
If you could even say that it is a fundamental right to have a right to algorithm transparency and protection against discrimination by algorithms. It is possible that AI might also affect the way the justice system works. The enforcement and application of the law in the absence of human empathy might become dangerous. I would not like us to over-centralize or to double various efforts in this field. Now from the point of view of our IMCo committee, I will focus on our consumers and our people, because a good healthy economy needs people, not just algorithms.
AI must support institutions, not replace them, must serve the citizen, not substitute them. AI must remain a tool, not a substitute for human decision. Digital transformation cannot mean a degradation of work conditions. Moreover, I would like to highlight how consumers can be misinformed with the use of the help of AI. We need to also protect against unfair competition that can benefit from AI.
And we can also, we should also focus on the risks of concentrating all databases in certain countries. In conclusion, AI should remain an instrument, not a substitute for human judgment. Europe should not have to choose between innovation and protection. Innovation means nothing if it does not improve the citizen's life. Thank you.
Sven Simon
Thank you very much. So now, we have shadow reporters who would like to take the floor, I think. And we start with a colleague from the Socialists and Democrats, Vithenes Povilas Antiochitis. Please.
Thank you.
Vytenis Povilas Andriukaitis
Thank you, colleagues. Thank you. Dear colleagues, first, I want to thank the rapporteur Emmanuel Kefalagiannis for very excellent cooperation and a very good report. This is a very important issue, and we have to have these discussions if we want to not be left behind, if we want our democracy to reflect the changing realities of the technology available, and if you want to harness or eat for the better. From the S and D group, I have just a few points to add.
In 2024, in Vilnius, the EU created the first legally binding international instrument on artificial intelligence. It includes a number of key concepts, such as a focus on human-centric AI consistent with human rights, democracy, and the rule of law, a risk-based approach, key principles for trustworthy AI, transparency for AI-generated content, support to safe innovation through regulatory sandboxes, and many more. We have to make sure it is reflected in this report. We have to especially underline the need for democratic and parliamentary oversight and accountability, human oversight, and review transparency as related safeguards as just our keynote speakers mentioned. It's a social area where AI can become prominent.
We have to ensure social fairness and workers' rights. Regarding treaty reform, I am supporting it in general, and I do believe we will have to include issues in treaties sooner or later. And I may support decision-making, but we have to make sure it can never replace political responsibility and never weaken parliamentary scrutiny. And our governments must reinforce representative democracy, but not replace it with technical, technocratic decision making. Of course, in any case, opening the treaties at the moment remains a problematic and sensitive subject.
We should include a call to all companies for a full implementation of an act and make sure that algorithm not interfere with democratic choice, which should be free, informed, and equal. In any case, thank you, and hope we will be able to negotiate a good deal together.
Ernő Schaller-Baross
Thank you. Thank you, sir. Mister chair, dear reporters, dear colleagues, first of all, I would like to thank the rapporteur for his work and for his openness throughout this long process. I particularly appreciate that prior to the presentation of the draft report, he engaged in several exchanges with the shadow reporters. This cooperative approach has significantly contributed to building trust and improving the quality of our work.
For me, the discussion on the institutional aspects of artificial intelligence is part of a broader and more fundamental debate. On the reform of European electoral rules, I have always been rather skeptical. And to be honest, I remain somewhat cautious today. However, I also recognize that our approach has become more open compared to previous discussions. I have revisited the issue, and it is clear that the current framework is not fully adequate to address the risk we are dealing with.
At the same time, we Patriots believe that we must be very careful. We are dealing with an exceptionally innovative and fast-evolving field, which requires restraint and a proportionate regulatory approach. We have recently adopted the AI digital omnibus package. The need of this omnibus process clearly showed that in this area, overregulation is the wrong direction. It risks not only stripping innovation, but also undermining Europe's own competitiveness.
I see a similar logic when it comes to the discussion on electoral law reform. At this stage, we should primarily map the available options and engage broadly with experts and all relevant stakeholders. I particularly welcome the fact that representatives of national parliaments are also part of this debate today. This is an important element of democratic legitimacy. The key point is that we should not rush headlong into complex institutional changes.
We should proceed step by step with due care and thorough consideration. I would apply the same principle to the question of a potential observatory. If we do decide to move in that direction, it should be strictly limited to an oversight and monitoring role, without strong executive or regulatory powers. From my side, I'm fully committed to the work ahead, and I look forward to continued cooperation with the rapporteur. Thank you for your attention.
Sven Simon
Thank you very much, George. Would you like to take the floor in your capacity again? Oh yes, Cheddar.
Sven Simon
Alright. Thank you. Then for renew, the Cheddar rapporteur, Helmut Panchetet, Peter.
Speaker 17
Thank you, chair. First of all, thank you, Manolis, Keffalo, Giannis, for your work, for your comprehensive report, and the clear presentation. We are living in an exciting yet challenging time. AI is set to affect every area of our lives, including legislative work. While we may need to change our work methods, we should not change our values, and I think that is the most important point for me.
Helmut Brandstätter
I'm also a member of the EUDS, the Democracy Shield, and it was mentioned here also the role of the algorithm. It's not the algorithm to decide what we want to read. It's not the algorithm to decide what we want. It's up to us to decide it, and that's very important here, especially in AI. Of course, we must remain open to technologies.
Speaker 17
We should not be naive or uncritical. Those new technologies, particularly ones as transformative as AI, will bring advantages and challenges. As elected lawmakers, we must always be aware that only we have the mandate to make laws, not machines. And one more thing which is important for me, it was interesting to listen to Vice President Commissioner Birckonen. She talked about 90 AI organizations, supercomputers, 8,000 startups, and so on.
But normally, you don't hear about it. We, as also the president said, it's always that the European Union also regulates. No. It's not true. This is a very innovative area, Europe.
Helmut Brandstätter
We are much better than we talk about sometimes. And for me, that's also important that we give not only the impression, and that's important that they are the members of so many parliaments here, to tell them what we are doing here in Brussels is for the good of the Europeans. And, of course, we need the laws because, and I stress it again, because we have the net values, but we also need technological sovereignty. And that's also the next point which I want to stress. Please also help us as far as technological sovereignty is concerned.
Thank.
Leoluca Orlando
You, chairman. Well, clearly, there's a requirement here to have a legally binding instrument. I mean, there are certain things that need to take primacy, though. Obviously, peoples, people, excuse me, rights and democracy. And I'm saying that because I think perhaps we should have a different way of expressing artificial intelligence.
I mean, obviously, the phrase artificial intelligence is widespread, and I'm fighting a lost battle here. But I prefer to talk about intelligence that is artificial, because the only real intelligence is human, whereas other kinds of intelligence would be artificial. But, obviously, I'm happy to use the phrase artificial intelligence, AI. But nevertheless, I do believe that we should ensure that people, right, citizens, and workers should come first. Now my second point is that we absolutely need to have European digital sovereignty, because that would guarantee that our values could be appropriately defended.
I mean, at the moment, everything's outside Europe. So we're not really looking for sovereignty here. It's more to have independence. Because at the moment, the competition is far greater abroad than it is in Europe. We're really not up there.
So I'd like to make a couple of comments on that. The first one relates to rights. There's a vast range of rights. And the second relates to freedom of democracy and electoral processes. And these are things that we'll have to pay appropriate attention to.
And I do think that the observatory needs to be a tool of control rather than looking at dealing with things that have already been dealt with upstream. And I apologize for having artificially extended my speaking time.
Sven Simon
From Italy was smiling a bit because the Italians and Spanish have a little more speaking time in the house. We know this is a cultural thing. So we'd no. But we don't have speakers from the left in ESN, so you could have take more time. Now I have been informed that a representative of the commission would like to take the floor.
The commission is represented by Thierry Boulanger, head of unit for artificial intelligence regulation and compliance. The floor is yours.
Speaker 24
Good afternoon. Thanks to the committee for inviting the commission to participate in this meeting. And thanks indeed for giving us the opportunity to also contribute to this exchange of views on the report, which is now being presented in this committee. As outlined in the beginning of the meeting by our Executive Vice President, we are working on making Europe an AI continent through, amongst other things, the AI content action plan, the Apply AI strategy as driving forces for the uptake of AI in Europe. And when doing this, when supporting our innovation, we also, in line with what we said here, want to make sure that this happens in a trustworthy way.
Speaker 23
And for this, as already underlined in several interventions, I mean, the main instrument is the AI Act, the first legally binding law on AI. We are currently working intensely on the effective implementation of this law, which provides legal certainty across the EU to ensure the development and use of AI happens in a safe way, in alignment with unions' fundamental values, financial values. The draft report, which was just presented, calls for several initiatives. For the sake of time, I will only comment on a few aspects. On the inter institutional dialogue on AI, we take note of the suggestion for a structured inter institutional dialogue between the Parliament, the Commission, and the Council on institutional implications of AI.
Speaker 25
In that context, it is important to mention the ongoing collaboration under the auspices of the working group established by the INCO and LIBE committees of the European Parliament on the implementation of the AI Act. This working group was set up in October 2024, so shortly after the adoption of the IAC, and it has already met more than 15 times since then, discussing many different aspects related to high policy and implementation of the regulation. On the also very important point related to international engagement and the possible need for a new AI convention at global level. We can only agree on the importance of cooperation at a global level on addressing challenges posed by the development and adoption of AI technologies. In collaboration with member states and European External Action Service, the commission is actively following the well-being carried out in different international fora.
Speaker 23
I could mention the OECD, the United Nations, but also the Council of Europe. Concerning the Council of Europe, we've been very actively directly involved in the negotiations of what remains so far the only international treaty on AI, so the Council of Europe Framework Convention on AI, sorry, also called the Vilnius Convention, which these negotiations were successfully concluded in 2024. And yet, important to be in mind that this convention was negotiated not only with the Council of Europe member states, but also non-European countries, including major players in AI. And so far, it was signed by the EU, but also 18 countries, including Canada, Japan, The United States. We also welcome the consent, which was recently given, okay, by the European Parliament to the article of this Treaty.
So here, just in line, I mean, echoing what we said about the time it takes to develop a treaty. We, rather than starting a new process, would really think it's important to encourage all the signatories of the convention to continue and finalize their process of ratification and to encourage Council of Europe member states to sign the convention if they have not yet done so. I would stop here because indeed, sorry for that, I went way over the time which was allocated. Thank you.
Sven Simon
You. Okay. Now we will follow with speakers from the European Parliament, and then we switch all the time, one speaker from the European Parliament and then two speakers from National Parliament, starting with Lauren Vincent from the EPP Group of the European Parliament.
Lauren Vincent
Thank you, chair. They are colleagues on behalf of the EPP Group. I would like to warmly thank our colleague, Emmanuel Cafaleanis, for his serious and committed work as well as for the inclusive approach throughout the drafting process. This report makes one point unmistakably clear: artificial intelligence is no longer merely a technological issue. It has become a constitutional and democratic question, at the very core of the European project.
Speaker 30
It is encouraging that the European Union and particularly the European Parliament are not reacting passively. At the same time, we must remain realistic and cautious. The European Union should avoid overregulation. Technological development, and AI is advancing at a pace that legislation can hardly match. If we regulate too hardly and too rigidly and too early, we risk stifling innovation, undermining competitivity. The key is smart and proportionate governance.
Lauren Vincent
In this regard, the proposal to establish the European Parliament's AI Observatory is a welcome step. It strengthens democratic oversight and expertise without unnecessarily constraining the innovation. We must also address one of the most pressing challenges, the impact of AI on democratic processes, particularly elections and public discourse. The growing presence of AI generated content raises serious concerns about manipulation, disinformation, and erosion of trust. Therefore, a fundamental principle of any regulatory framework must be clear. We must always be able to identify human responsibility behind AI-generated content and actions.
Accountability cannot be outsourced to the algorithms. At the same time, we should not overlook the positive potential of AI. When used responsibly, it can improve legislative quality, support better implementation of EU law, and enhance coordination among the member states. And probably, we will hear such stories or ideas from representatives of the national parliaments. Finally, the report rightly positions the EU as a global standard setter.
Speaker 28
Europe should aim not only to regulate internally, but also to shape international norms while remaining open, competitive, and innovation-friendly. Thank you.
Ricardo CARVALHO
Thank you very much, chairman, colleagues. Well, we live in a historic time. AI has gone from being a distant promise to a transforming and frightening force. It's having deep-ranging effects on production, our economies, on our democracies, on the daily lives of our citizens, of our states, and also on governance. And given the situation, the question is not whether we should regulate it more or implement, but we should actually implement what has been regulated.
I mean, it's not really an issue. Everyone knows the answer to that question. The real question, though, is how can we effectively do all of this in our interconnected world? The response can be European, at least partially, but actually, that's not enough. We need a global response.
Let me, we have a regulatory framework that's ambitious and based on values, but AI is decentralized, transnational. It's constantly developing, and it requires an approach that goes further than borders. It needs to be cross-continent in nature because otherwise, we risk fragmentary regulation, and we also risk not having appropriate controls on various risky uses of AI. And there may be a problem in terms of railways, water, gas and other essential services. I mean, there have been recent impacts in the world, such as COVID, forest fires, floods, etcetera.
And all these things could be helped by AI, perhaps. Now we need to be ambitious, but we need to work appropriately. We can't do things too quickly. So what can we do here? How's the European Union going to be able to deal with its differences and come up with an appropriate strategy?
That's the big question, I think. Thank you.
Thank.
Céline Calvez
You, chair. Too often, innovation and regulation are presented as being connected, and the proof is the adoption of the AI Act in 2024. We were the first entity to be able to politically and legally innovate by proposing the first piece of legislation on AI, but its implementation has been dissatisfying given the fact that we have challenges relating to fundamental values of the EU. And the rapporteur mentioned that we AI must remain aligned with the EU values. Let's look at copywriting.
The recent adoption by the EU European Parliament of Axel Vos' in NE, that there was a vast majority there. That report was looking at copyrighted material and how that might be used by AI systems. This report wanted to provide legal certainty to AI services and to protect the copyright holders. Among the different recommendations, there should be some kind of trust-based mechanism between these two entities, and there should be rules relating to training AI systems. There is a French law that was just adopted by the senate at first reading relating to this.
We recognized the compatibility of this law and the EU AI Act from 2024. Our ambitions when it comes to sovereignty and innovation are strong and will not be countered, but this now probably presents the opportunity to recognize copyrighted material as EU heritage. This is a strategic, intangible asset, and this is not something that should be endangered. AI should allow us to better protect our heritage as a European heritage.
Sven Simon
I will before I give the floor to Brando Benify, I will hand over to the chair, the chair to my dear colleague, vice chair of the committee, Gabriela Bischoff, as my further obligations require me to leave now. Thank you for having been here in the European Parliament, and go on. And the next speaker, but don't apply this rule with the Italians, can speak longer. No. No.
Brando Benifei
Thank you very much, Chair. Let me also thank the rapporteur, our colleagues, and all the guests of this valuable inter-parliamentary committee meeting. I'm personally very interested in this discussion, also in light of my different roles in this parliament as co-rapporteur of the AI Act, co-chair of the working group on the implementation and enforcement of the AI Act, and member of both AFCO and IMCO committees. Let me congratulate the rapporteur, Emmanuel Kefaloganis, for the work done. I believe it's a balanced text.
Let me mention some key elements. First, the stronger emphasis on a human-centered approach to AI governance, ensuring that innovation remains aligned with fundamental rights, the rule of law, and democratic values. Second, the focus on democratic stability, recognizing the growing impact of AI on political communication, electoral processes, and public trust, and the need to safeguard the integrity of our democratic systems. Third, the attention to the relationship between private and public governance. While private actors play an important role, democratic institutions must remain responsible for setting the rules, ensuring accountability, and protecting the public interest.
We can even do more. I would underline how AI and digital technology can strengthen democracy participation, improve decision making, and bring citizens closer to new institutions. The challenge is to make participation more accessible. In this context, let me introduce the concept of civic AI, using AI to support citizen participation, which has strong potential, provided risks such as bias and lack of transparency are addressed. Second point I want to raise is that I think that the proposed European Parliament AI observatories are an interesting idea to strengthen parliament analytical capacity.
It's very important that it integrates with existing bodies like the AI working group and the AI office. So, working on the administrative part rather than the political sector. So I, to conclude, I will be happy to contribute constructively to the amendment phase of this report with the aim of enriching the valuable work already done so far, together with the shadow rapporteurs, including the one from my political group, mister Andrew Kitis. Thank you very much.
Gabriele Bischoff
Thank you very much, Brando. And the next speaker is Miss Greed Van Reichegem. I hope I say it right.
Greed Van Reichegem
Thank you, Missus Chair, colleagues. Thank you for organizing this, committee. I understand European integration as a process of ever-closer cooperation between member states, aimed not only at strengthening the internal markets in economic terms, but also at deepening institutional and societal alignment. The European Union has already positioned itself strongly through legislation. While a clear framework is indeed necessary, it should enable innovation rather than close doors.
With regard to regulation and oversight, the EU AI Act introduces strict transparency obligations from August 2024, particularly for generative AI systems. Another key challenge concerns algorithmic decision-making in the public sector. It is essential to guarantee meaningful human oversight and to prevent opaque decision-making processes that disregard the human dimension. Protection of democracy is equally critical. AI can be misused to spread disinformation and manipulate electoral processes.
The integrity of the information space itself is at stake. Greater emphasis should be placed on verifiable facts rather than opinions. We must also address the growing digital divide and the lack of internal expertise. There's a real risk of fragmentation and lack of direction due to insufficient AI knowledge within parliamentary institutions. Finally, accountability and transparency remain fundamental. Parliaments must retain control over algorithmic processes and ensure that AI systems are understandable and auditable.
AI presents opportunities across all sectors, but to seize them, we need a solid understanding of what AI is, how it works, and how it can be applied. In other words, to conclude, I have heard important keywords as democracy, cooperation, trust, smart, strategy, infrastructure, and investments. AI is not only a regulatory challenge, but a strategic opportunity for European integration. Artificial intelligence needs intelligent institutional cooperation in Europe. Thank you.
Gabriele Bischoff
Thank you very much. And the next speaker is Miss Elizabetta Gardini.
Elisabetta Gardini
Thank you, madam chair. Well, me begin by thanking Mr. Kefalo Giannis for his excellent work. I very much enjoyed reading it. And it really does have a very clear positive point, because it really focuses on the institutional and constitutional side of AI rather than the regulatory and technical side.
And that's what we need to be looking at, because now AI is no longer just a tool. It's something that affects the way in which we build consensus, the quality of information, and also the balance between technical power and political responsibility. I mean, it basically influences democracy, as we've heard. And this is something that's key, I think. So the point is not just how we should regulate AI, which is something that I think sometimes we focus on too much.
The point is who's going to regulate it and who's responsible. I don't have a lot of speaking time, so I want to focus on the role of national parliaments, because it's all very well to boost the analytical capacity of the European Parliament. But I don't think there should be too much overlap or some kind of centralization, because national parliaments aren't secondary parliaments. They are places where there is direct democratic responsibility, and there's a great deal they can do. I mean, the Italian parliament worked hard on AI.
So I think that's why it's important to have this inter-parliamentary cooperation. And it really needs to be clear, as in we need to be discussing our experiences, exchanging good practices, etc. It shouldn't just be, it should be a debate discussion that takes place in terms of substance rather than form. So again, as we've heard, we need to ensure that we look at the human side, because algorithms aren't neutral, because those who are responsible for algorithms control the information and control what comes out. So we need to govern what's happening.
Gabriele Bischoff
The word to my dear colleague, Charles Gurens. I just would like to remind you that we still have 8 speakers. We are already a bit behind schedule, so I encourage you to stick to the two minutes. Or even if you need less, I think it's very much appreciated.
Thank.
Charles Goerens
Thank you very much, madam chair. Let me reassure you straight away, I'm not going to exceed my speaking time. I wanted to highlight something. The EU was in fact the first to take a decision whereby we would systematically have a connection, a link between AI and human responsibility, or indeed political responsibility. So that's an angle, if you like, on AI, and it's a priority for many of us.
And it's to make sure that AI is at the service of our societies. The EU was the first to understand this, to comprehend this, the first one to be determined to legislate in this particular area. In a world, of course, where might is often seen as right, nevertheless, we wanted to make sure that the European Union is recognized for the fact that it had that clear sightedness and that it came forward with the legislation that we're all familiar with now and which I think sets us out. Everybody, and in particular across the Atlantic, you know, not everybody else sees it the same way as us. It has to be said.
And this is going to be a constant point of tension or test for us to maintain our legislation, to defend our legislation against attacks coming from elsewhere. You know, this is going to be constant testing of our system, and I hope that we'll be resilient enough, strong enough. And I very much hope that ethically, we will also stand strong here in this house, and also in our member states. There are attacks coming from elsewhere. We need to stand firm.
We shouldn't be discouraged. In fact, we should make sure that our laws, our regulations in this particular field become, if not global standards, at least international standards when it comes to implementation. Thank you.
Pavic
Thank you very much. Dear colleagues, it's a great honor to address you today, not only as a member of Croatian parliament, but in my capacity as vice president of parliamentary assembly of council of Europe. We are discussing today the draft report, and thank you and the rapporteur for this opportunity to exchange the views. I wish to complement the report and underline the Council of Europe framework in this field, meaning the Council of Europe Framework Convention on Artificial Intelligence and Human Rights, Democracy, and Rule of Law that was signed on 5 September 2024. The convention represents a historic milestone.
For the first time, we have a global legal framework assuring that the entire life cycle of artificial intelligence systems is fully consistent with human rights, democracy, and the rule of law. It clearly establishes that AI must remain human-centric, grounded in human dignity, individual autonomy, transparency, and accountability. It also introduces obligation for effective oversight, risk assessment, and accessible remedies for citizens affected by AI systems. Crucially, the convention places strong emphasis on the protection of democratic processes. It requires that AI must not be used to undermine the integrity of elections, democratic institutions, or public debate.
It safeguards citizens' ability to form free opinions and to participate meaningfully in democratic life. In this context, the role of parliaments is absolutely essential. As highlighted in this draft report, the European Parliament bears a specific responsibility to ensure democratic scrutiny, transparency, and accountability in the deployment of AI technologies, especially where there affected electoral processes and public decision making. The same applies to the national parliaments and us in the national parliaments. Together, we must assure that technological transformation strengthens rather than weakens democratic legitimacy.
In this constant context, the European Union has taken a decisive step forward with the artificial intelligence act. It reflects a clear European ambition. From my perspective as the Croatian parliament and from my role of PACE vice president, I strongly support enhanced cooperation between national parliaments, the European parliament, and international organizations as this report states. Thank you very much.
Gabriele Bischoff
Thank you very much. And next speaker, Mister Theodoros Skylakakis from Greece.
Theodoros Skylakakis
Thank you, madam president. This excellent report by colleague, dear colleague and friend Manolo Skifalo Janis gives me an opportunity to make an AI wake-up call. With AI, we don't have a new technological tool. We are at the beginning of an economic and social revolution more transformative than all the industrial revolutions we have faced up till now. The difference is that it is coming five to 10 times faster than industrial revolutions, and it will be five to 10 times more important.
This brings us challenges and opportunities much broader than we currently discuss. I will give you two examples. One challenge. 20 to 30% of our GDP is produced by knowledge workers, accountants, lawyers, architects, doctors, etc., working in inelastic nontradable services markets. AI will increase explosively their productivity in the next 5 years.
The demand for their services will not increase accordingly. And if we don't have policies to manage the way AI enters these markets, we will have prices dropping. We will have incomes collapsing in big parts of our economy. Opportunity, public procurement in the European Union, 2,500,000,000,000 per year. Most of it, 40%, I would say, goes to one bidder, one way or another.
Introducing AI there could save us maybe 100, maybe 200, maybe €300,000,000 per year. So the opportunities and the challenges are much broader than we are currently discussing, and we must make this discussion a lot more efficient and a lot bigger than we currently have. Thank you.
Gabriele Bischoff
Thank you very much. And the next speaker is, from the European Parliament, Mr. Wiegheny.
Wiegheny
Thank you, Chair, dear colleagues. I like to congratulate the AFCO leadership for this exchange. It's an excellent idea to bring together national parliamentarians and members of the European Parliament because although the II Act was adopted here in the European Parliament, it is our common responsibility because implementation then is rather in the hands of the national member states. I appreciate the draft report of my colleague, Foloyanis. It focuses on a different aspect of this AI revolution.
If I can summarize on the democratic oversight of that and how can we guarantee that. And since we don't have much time, I would like to focus on one aspect here. The EC vice president, Virkonen, mentioned the AI office and the fact that the employees there are 140 at this stage. And I can assure you that it's already not enough. The AI office is not able to produce the necessary guidelines in time.
And the more I am dealing with the AI and the AI regulation, AI oversight, I am more and more convinced also as a rapporteur of the socialists on the AI and the digital omnibus and the AI act regulation, that we need a more comprehensive oversight infrastructure or structure. I believe that we, at this stage, already need an AI agency that would be more independent. At the same time, would make it easier for European Parliament to oversight to scrutinize the activity of such an agency. And we also need a more centralized approach in the member states. I give you only one example.
For example, on the issue of fundamental rights, there are seven Bulgarian institutions that have been tasked to follow this. I think this is not the way to deal with such a difficult topic. So let's think about that. I'm sure we'll come to this issue sooner or later. At this stage, I personally support the idea of such AI observatory in the European Parliament.
Thank you.
Thank.
Antonino Iaria
you. Thank you, Rapporteur. Thank you, Chair. Well, you've all said, knowledge is power. That's the case now, more than ever.
And the knowledge of AI is power as well. So the question is, how can we distribute this knowledge so that it can be more democratic? Because having this power in the hands of the few, those who know how to use it, those who manage it, those who produce it, is dangerous for democracy, as we've heard. We in the Five Star Movement want to see investment in digital literacy, and we want to see education at all levels so that people can know about AI, but also have the capacity for critical thinking, so that people can take advantage of opportunities and also use AI effectively. So let me conclude just by saying that investment cannot only be in data centers and computer capacity.
We also need to invest in diffusing knowledge, because that is what could make AI more democratic. And let me conclude now just to do away with the myth that Italians speak at too great length. Thank you.
Gabriele Bischoff
Absolutely. That is very much appreciated. And, also because some messages transport better if they are shorter sometimes. But that's my own conviction. Conviction.
Next speaker is Patricia, Madame Patricia Gonzalez from Portugal.
Thank.
Speaker 43
You, madam chair. Well, AI is often presented as a technological opportunity but also as a regulatory challenge, but we should recognize what it actually is. It is something that has significant externalities in terms of the environment, geopolitics, etc., and these can't be dealt with in isolation. Much of what we do in terms of governance at the moment continues to focus on the direct impacts on users, but the wider effects are even more difficult to regulate. There's a rapid expansion of AI infrastructure, and that means they use more and more critical raw materials, water, and energy.
These resources aren't distributed in a fair manner, and the costs are often borne outside the jurisdictions where AI is developed and used. At the same time, we see an increasing concentration of power in the hands of just a few global players who control data, computer capacity, etc. Now this isn't only an economic issue. This is an issue of sovereignty, responsibility, and democratic resilience as well. These are issues that can't be dealt with just via national regulation.
We need international cooperation, which goes further than just harmonizing rules. We need to look at who is responsible. The issue is not just how we regulate AI within our borders, but how we can guarantee that its global development is sustainable, takes place in a responsible manner, and is aligned with our values. If the EU wants to lead in this field, it also needs to lead in building an international cooperation framework that deals with these externalities, not as secondary effects, but as key concerns. Thank you.
Gabriele Bischoff
Thank you. And our next speaker is also from the European Parliament, Madam Gaziuk Pihovic.
Gaziuk Pihovic
Mhmm. Thank you for organizing this debate at the European Parliament. It's obvious that whenever we want to adopt any regulations, the approval of various political forces will be required. So I will talk about the political aspects here. Elon Musk's GLOG generated 3,000,000 naked pictures of real-life people within a few days, over 100 images per minute.
Speaker 45
We're talking about AI-generated pictures of real-life people without their consent. 23,000 of those pictures were pictures of children, some which were child pornography, and there were also ordinary women who were the victims of these activities. They discovered their naked pictures online. And politicians also protest against any regulations of AI, against the AI Act, DSA, DMA. And everyone at the right wing talks about censorship, why we only want to protect women and children.
Speaker 47
We want to have a minimum of dignity and respect online. Mr. Bosak from the Confederation political party in Poland was proud of Hitler. He said, welcome aboard. We are on board together, fighting political correctness.
Speaker 46
So that's the right-wing parties. They fight against democracy, against political correctness, whenever they can gain something. So let us ask this question to the right-wing parties. Are these your conservative values? Is this the protection of the families that you are talking about?
Gaziuk Pihovic
Is this the protection of women's dignity? Millions of naked pictures of children generated by artificial intelligence. And the right-wing parties talk about censorship while Europe wants to stop it. It's a shame. What is our approach to artificial intelligence?
Speaker 46
Will we treat it as a global problem, as a systemic problem? This will show whether we remain loyal to our European identity, which stems from Christian roots, from general human values. So we need to do our best for the artificial intelligence to be used to the benefit of people, and not to harm those who are weaker. Thank you.
Susana FERREIRA GONCALVES
Thank you very much, Madam Chair. I'd like to congratulate the members of the European Parliament and thank the rapporteur for his work. Now I think AI is a factor that is going to affect the way that we take decisions as elected representatives, and how citizens relate to the state. We need to develop a more local perspective. We need to see more decentralization, and we need to boost the powers of local authorities.
But we want to see the practical use of AI taking place at the local authority level: health, mobility, the management of public services, public tendering, and administrative management, etcetera. And there's a clear challenge here, which I believe is very important. The implementation and the use of AI can't simply be limited to the central administration. So there are a couple of issues that I want to raise in that context. How can we ensure that the capacity building, cooperation, and scrutiny mechanisms effectively include local and regional authorities, where many of these automatized decisions are going to be implemented?
And how can we ensure that this future AI observatory can be a useful tool for members of national parliaments to accompany this multilevel implementation. Now our technological independence, but it will only work if we have territorial cohesion. We need better public services without compromising the human nature of developments and ensuring that we don't have too great inequality. Thank you.
Gabriele Bischoff
Much, for all your input here and the exchange. I think it shows also that both sides can profit from the European Parliament, US members of national parliaments to focus concretely on a report like this. Let me inform you, and especially the colleagues from the European Parliament, that the deadline for amendments is 04/28/2026 at noon. And, I think after this fruitful exchange here, we will go to a new panel on parliamentary cooperation in AI. And it's, I would like also to invite mister Thierry Sauter, vice chair of the committee on European affairs of the French National Assembly, to join us here on the podium for the first part.
Yes. And maybe I already invite Rina Sikud here for the second panel, then we are complete here and can start going into our debates here. So we have discussed it already: artificial intelligence. We all know it. AI is transforming not only policies, but also how parliaments function, how they legislate, and also exercise scrutiny.
And, as it is highlighted in the draft report here, strengthening cooperation between the European Parliament and national parliaments is really essential to reduce what was also addressed before in the keynote speech, to reduce information asymmetries, to build shared expertise that was very much emphasized all over, and to ensure coherent democratic oversight across the union. This is also closely linked to the idea that we discussed.
Of enhancing parliament's organizational capacity, including through a dedicated analytical structure supporting all committees here in the House. And I'm very pleased.
Linked here to have Mister Thierry Sauter here as vice chair of the committee on European affairs of the French National Assembly. Your experience here at the national level is really essential to building a real multilevel parliamentary approach to AI governance. And, I invite you to deliver your remarks here, Mister Souti, for three minutes. We might have a little bit more time because the second speaker could not be here. So the floor is yours.
Thierry Sother
Thank you very much, ladies and gentlemen, members of the European Parliament, dear colleagues, members of national parliaments. We're talking a lot about AI as being a technical, industrial, and economic challenge, but I think we need to spell out clearly that above all, AI is a democratic challenge, because behind the algorithms, behind the models, behind these promises of efficiency, there is a key question. Who's deciding on this on the basis of which rules? What transparency is there? And where is the oversight in this?
That's really at the heart of the issue. We have experienced huge technological revolutions over the course of history, and they have all changed political relations, how our institutions function, and public life. AI will be no exception to this rule, but it is doing so at an unprecedented speed, scale, and we as parliamentarians have a particular role to play here. Because if we don't fully understand and address this revolution, then other people will decide for us. Markets, huge platforms, private stakeholders, the most powerful people, and systems will perhaps remain opaque in how they work with their algorithms.
And so, therefore, as parliaments, we cannot run after AI behind it. We need to take a firm stance and grab the bull by the horns. We need to have harmonization because there's too great fragmentation of our parliamentary reports, parliamentary relations. And this cross-cutting challenge posed by AI needs a harmonized approach. We do have useful work that has been done, but we don't have the necessary tools and structures that would allow us to carry out a political scrutiny of this challenge.
And yet AI doesn't fall into just one box. It affects public freedoms, culture, education, economies, security, information, work, public services, and even how citizens are educated. It is a cross-cutting issue that affects all aspects of society, and therefore, needs a reform response. But there's a second aspect. No national parliament can tackle this on its own.
We need to set up methods of cooperation. We need to pull our forces. Our markets, our democracies around Europe are all exposed to the same risk, and our citizens are all facing the same phenomena: information that's been shared, content, automatic production of text and images. And as a result, we're all working with the same invisible friend, the algorithm that decides what we see. AI dictates what social media shows us.
So we cannot work in silos, given that this technology is sprinting at a global level, making leaps and bounds forward. We need more regulated interparliamentary cooperation, and that's something that we can talk about here over the next few days, in Brussels. This needs to be more structured, and it needs to be workable. There are fora for sharing of best practices, but we need to identify blind spots and share challenges and problems that we're facing. Shortcomings are important to discuss because AI and algorithms in particular are evolving rapidly and the offer on the market, the products out there are changing very, very quickly.
And we want to be more agile in our response at the national level and at the European level. We have learned by experience that we need to correct our course, and this should be done through discourse between our national parliaments. Europe cannot be a continent that looks back on technological changes after they've happened and regulates them, but we need to be the continent that really stands in the face of certain practices and condemns those practices. We need to protect our values. Our line is clear.
Innovate. Yes. Accelerate. Yes. But we must never abdicate our democratic duties.
This is part of our parliamentary work. We're seeing our analysis, ways of working, and our interactions with citizens. We need to better understand consultations, petitions that are coming to us. We need to group this all together and address the issues. We should see the AI as an opportunity for democratic renewal.
The challenge is not just about regulating a technology, but really what is at stake here is working at whether European democracy can work together transparently and responsibly, to the benefit of the people, or whether we will slowly slip towards a situation where we are no longer able to control systems. I think we need to make a clear choice here. We need to recognize the power of innovation, but we need to support and guide that innovation. Democratic oversight, however, must remain a key asset at national and European level to guide that regulation and to build and support AI. Thank you very much.
Gabriele Bischoff
Thank you very much for giving a lot of fruit and input for the debate here and experience, and you will have one and a half minutes each. And we start with Miss Domenica Spinelli from Italy.
Domenica Spinelli
Thank you, madam chair. Well, as we've already heard, the adoption of AI raises important questions in terms of responsibility, safety, security, excuse me, and the safeguarding of democratic processes. At the same time, there are opportunities provided also in terms of the potential improvement of interactions between citizens and institutions, and that requires a boosting of the parliament's role. As Mr. Gill said earlier, we need major regulation of AI, but it needs also to be flexible and not too centralized.
Speaker 50
These are complex issues that can only be dealt with via structured cooperation, with constant work being carried out by the European and also the national level together. We need to be focusing on major strategic challenges, as we've just heard from Mr. Souté. The working group on the implementation of AI is a useful example. As we saw in a working group in Cyprus, we need fruitful exchanges between all colleagues in institutions, and we also need fruitful cooperation with the regulatory authorities and member states, because it's only thus that Europe can be involved in this challenge.
As Miss Roberta Metzola said earlier, we can become the next Silicon Valley in AI. Thank you.
Sir, president.
Apologies.
I didn't want to take the floor, actually. Thank you.
Arash Wili
Yes. Thank you. Honorable chair, distinguished colleagues. On behalf of the Alliance of Voivodino Hungarians, it is an honor to address this committee. Since 1996, our party has been the sole parliamentary representative of the Hungarian community in Serbia, consistently working towards our communities, European integration, and the future.
As we discuss the institutional aspects of artificial intelligence today, I am proud to announce that the Hungarian community in Boivadina is not merely observing these changes, but actively shaping its own future. At the initiative of our president, Doctor. Balin Pastor, we have developed the artificial intelligence strategy of the Hungarian community in Voivodina for the 2026 to 2033 period. The process began in November 2024 in Subotica. We recognize that AI will fundamentally define the future of education and the economy.
Our goal is clear. Our community must emerge as a winner of this technological shift. Our strategy is the result of the dedicated work of 14 experts and 6 reviewers. We focus on key sectors such as public administration, health care, agriculture, and minority language rights. The latter is of paramount importance to us in ensuring digital equality.
This strategy is our guarantee that our community will remain strong, competitive and forward looking within the European space in the age of artificial intelligence. Thank you for your attention.
Gabriele Bischoff
Thank you. And the next speaker is Mr. Alexander Kriebitz from the Interparliamentary Union.
Alexander Kriebitz
Madam chair, honorable Rebetur, honorable members of the European and National Parliaments, ladies and gentlemen, thank you for giving me the floor. It is an honor to address you today on behalf of.
Alexander Kriebitz
Interparliamentary Union. The IPU brings together more than 180 parliaments worldwide and is grounded in a simple yet powerful conviction that institutional cooperation and parliamentary dialogue are essential to advancing democracy, human rights, and the rule of law. Parliaments have a dual task, legislating effectively and holding the executive branch accountable for the enforcement of those laws, but also their interpretation. AI is now testing these very foundations, as outlined in the report. The pace of change often outstrips deliberative democratic processes and makes political decision-making less transparent and harder to supervise.
The IPU is taking decisive steps to support parliamentary action on AI. We are observing high levels of interest among parliaments and many emerging parliamentary initiatives on global AI governance. Part of our role is to help parliaments share lessons learned and good practice, in doing so to contribute to the growing consensus for political action on AI. The 2024 resolution on the impact of AI on democracy, human rights, and rule of law, as well as the 2025 Kuala Lumpur declaration on Parliament's and Responsible AI, reflect an increasing parliamentary awareness globally on AI's impact on society. The declaration sets out concrete recommendations for parliamentary action on AI policy, and parliaments involved in the declaration are called upon to provide a progress update by June 2026.
AI is also a central theme of our 152nd assembly taking place in Istanbul right now. We view today's discussion in the European Parliament as essential, given its pioneering role in AI legislation oversight. Because parliaments act as the bridge between constituents and governance, inter parliamentary cooperation is the most effective way to ensure that emerging frameworks on global AI governance are, by design, embedded in democratic processes and attentive to different regional, economic and developmental realities in the global North and South. Let me conclude with a final remark. Governing AI is not merely a technical issue. It is, at its core, a democratic and parliamentary responsibility.
It requires inter-parliamentary collaboration and the involvement of parliaments early on in ongoing dialogues on global air governance. It is one area that parliaments must shape together. Thank you.
Shivota Stasiewicz
Distinguished chair, honorable colleagues, discussion on institutional aspects of AI comes at a pivotal moment when AI seems to be merely a technological tool and becomes a cornerstone of modern constitutional order. As noted in the report, it is key to democratic legitimacy and rule of law. Serbia enters this dialogue from 39th position globally according to the latest government AI Readiness Index for 2025 by Oxford Insights. This ranks us among the top 20% of the world's most prepared economies. Our ambition is further codified in our national strategy 2035.
Central to this vision is the development of our domestic large language model, LLM. Serbian e-government aim to revolutionize public administration, agriculture, and health care. We anticipate the automation of our economy, supported by deep reforms of our education system to prepare our citizens for this new reality. Within the framework of horizontal parametric operation, I can stress, Serbian High Readiness qualifies us to be a committed partner in the future European parliamentary AI observatory, offering our expertise for a coherent algorithmic oversight across the continent. We are ready to develop joint mechanisms to protect our democracies against deepfakes and disinformation.
Serbia wishes to be a part of new constitutional code within the EU treaty changes, ensuring that it strengthens rather than weakens our democratic institution. Our strategic goals, our global ranking, confirmed that Serbia has the capacity and the will to be a leader in digital integration. Our message is clear. We are building the future where technology serves the person, and the parliament remains the supreme guardian of that responsibility. Thank you.
Gabriele Bischoff
Thank you very much. And the next speaker is Mr. Antonio Iaria from Italy.
Thank.
Antonino Iaria
you. I know it's pronounced Iaria. Well, we're talking about cooperation between national parliaments and the European Parliament. So I think we need to try and find a way of establishing a common policy in terms of governing AI. I mean, I'd quite like to ask AI the question.
But if the AI in question hadn't been trained with what we've produced in our national parliaments in the past, it may well give an answer that would be biased towards the European Parliament. So I think we really need to manage this tool and create the only sandbox that we could possibly create rapidly by trying to find a balance between policies at the central level and also those from the national parliaments? Thank you.
Gabriele Bischoff
Sir, I have still one speaker on my list, Mr. Theodoros Skylakakis.
I will learn it.
Theodoros Skylakakis
Well, thank you, madam chair. I would start by saying that, I'm listening that we are governing AI. I'm afraid that before governing, we need to understand AI and understand what's coming. We have three waves that we're going to face in the years to come. The first wave will be AI in knowledge workers.
It's already happening. Then we will have automation and robotics powered by AI. It will come five, ten, fifteen years from now, because it needs more time and more investment. And then there is a race that's not in our hands between China and America for general intelligence, artificial general intelligence, which is going to change humanity and the social contract. And we don't know when this will occur.
Some people say in 5, some in 10, some in 15 years. What we need to do as parliaments is to work together to bring the knowledge of what's coming first, because every policy that we are going to implement in the next 10 years will be totally dependent on this transformation.
Gabriele Bischoff
Indeed. Thank you very much, and everyone for your participation and also for highlighting that it is a democratic parliamentary responsibility that we have to address together. And also as the introductory speaker, missus Sauter, already said, we have to reflect on this exchange between European and national parliament in a more structured and also sort of ongoing way, and to take this with us because it will help us on both sides here to address it. And we will go to panel two now. And a key question, that was raised also on democratic safeguards in the European Union, electoral integrity, and institutional evolution.
That is very important. And, because as we know that AI is increasingly affecting the conditions under which democratic processes are organized and they operate, and this also refers very much to the core to, elections. So from an ASCO perspective here as a committee, this raises for us some fundamental questions. And the key question is really the resilience of our electoral systems. Key question to address, the adequacy of the current legal framework here, but also the broader impact on democratic legitimacy and the institutional balance.
So, what we can see here also, and we looked at the work we have done, we can see that the European electoral framework is still largely fragmented, and it might not be sufficiently equipped to address AI related challenges. And this may call also for further harmonization under article 223. But more broadly, AI in the longer term might also require reflection on the evolution of the treaties here, really to preserve democratic accountability and also institutional coherence. And I'm very pleased here to have Miss Rina Sikut here, a member of the finance committee of the Estonian parliament. And, if I might emphasize this too, we worked together very well with the National Parliament and the European Parliament in the Conference on the Future of Europe, and she was chairing the Digital Working Group there.
So, Rina, you have the floor.
Rina Ronja Kari
Thank you, madam chair. And it has been a very interesting debate. And I would paraphrase president Metzola, who compared the US, China, and EU approach by listening to your statements that China uses AI to control people. We believe that US lets AI out of control, but EU keeps humans in control. And I sense that we have a very strong consensus on that.
But a lot has been said about the risks, about disinformation, deepfake campaigns, how the political communication has changed, and this is all accurate. This actually happens. It already affects the way we work, the way election results are determined. But at the same time, we have to admit that AI is not somewhat, I don't know, meant or better fit for foreign actors to intervene in elections in EU member states or better fit for disinformation campaigns. Since one of the prerequisites of democracy is citizen participation and an informed citizen.
We have to also discuss the ways we should achieve, as my colleague Mr Soda said, democratic renewal. We can use AI to better engage people. I don't know. We can have digital mini publics to deliberate policy issues with citizens in member states, but also on EU level, real-time debates in different languages. We should make use of the technology to bring the EU closer to citizens and then member states' policy issues, also to the dinner tables.
For example, we can use AI to let people ask how this specific regulation, whether it's a directive or a change of law in a member state, affects their life, their daily life. There are new ways we can come up with to improve the way democracy works in Europe. We shouldn't let AI be used for, I don't know, risky activities or unwanted activities. So I think this is a common issue for us as well. No one knows the solution in Estonia.
We are proud to be a digital state. We don't know the answer. Our people's trust towards political parties, politicians, members of parliament is at an all-time low level. But AI currently is eroding trust. We It's up to us to turn this around.
And although I'm used to, I don't know, TV, policy debates, and opinion pieces in newspapers, we also have to admit that AI and, I don't know, social media lives, memes that are shared might bring more younger politicians or different types of politicians to engage with people and explain policy issues to them the way they prefer it to be done. It is not a publicistic way, but a fact-based, evidence-based way that is available behind the dinner table or on their social media feed. So in addition to all the risks that have been described here, I would also welcome you to discuss or debate on what should be the institutional evolution to improve the way democracy works in Europe and how do we achieve the democratic renewal that was mentioned? Thank you.
Gabriele Bischoff
Thank you very much. And, the floor is now to the members of the European and the National Parliaments, and we start with my parliament colleague, Lopez Aguilla.
Juan Fernando López Aguilar
Thank you, chair. At this stage of our gathering of this interparliamentary conference, it's been much told about the impact of artificial intelligence on democratic processes. Certainly, artificial intelligence is not a threat. Artificial intelligence, it's a light-speed revolution which is underway, and it's a fact. So a negative reactionary attitude against artificial intelligence is certainly a no-go, certainly not.
But the thing is that that doesn't mean that we are to ignore that artificial intelligence is not neutral. It can actually damage democratic processes instead of fostering them. Artificial intelligence can actually serve democratic enhancement instead of hurting or distorting the democratic processes. And that's the point. We should be agreeing in this conversation.
What's the role of parliaments? Of course, consistent with the role that the European Parliament has exercised in the past mandate, putting in place legislation in order precisely to get hold of and tackle the challenges of artificial intelligence, making them compatible with human oversight, with accountability, with transparency, and hopefully, certainly, legal responsibility of the platforms behind them. But having said this, there's much to be done when it comes to harmful practices, dark patterns, enhancing digital literacy, training European citizens, educating European citizens to master the tools of artificial intelligence, and, of course, preventing interference of algorithms, particularly addictive algorithms and algorithm systems as to parliamentary and democratic processes. And that is the point of article 223, which is the one regarding the common safeguards that the European Union should be putting in place when adapting the European Electoral Act and its implementation into the national legal systems according to their constitutional requirements. So the final conclusion is that it's only obvious that strengthening cooperation, both of us, national parliaments, and European Parliament, is the challenge ahead in order to advance multilateral efforts grounded on democratic values and protection of fundamental rights.
That is the point. We should be certain in the forthcoming future when it comes to legally mastering artificial intelligence related challenges.
Irina Fekcikova
Thank you very much. It was very well pronounced. I just would like to add some very few things. And, one of them is that, actually, the fake news and deepfakes are actually meaning that there is a fundamental breakdown of the human, information ecosystem, and it has definitely consequences for democracy. We know quite well that lies are spread 6 times more than the facts.
When actually the X, formerly Twitter, enabled a feature to allocate the influential profiles of MAGA, it was seen that, actually, these MAGA accounts are located in Nigeria, Eastern Europe, in Southeast Asia. And that's very, very serious because we have an army of automated generated bots and trolls that are interfering with a democratic discussion of ordinary citizens and human. And, it allows to hack a domestic conversation and change a lot. And I see it as a real threat to democracy because freedom of expression is a human right, not an AI right, and this is something we need to bear in mind. And if we allow bots to speak on our behalf, we are actually violating the fundamental right to be heard of citizens and humans.
Speaker 59
And, therefore, we must close the bot loophole to protect our democracy. When we allowed 100,000 automatic accounts to flood our airwaves, we are not protecting free speech. We are allowing a thousand real human citizens to be silenced by artificial crowd.
Gabriele Bischoff
Thank you very much. And, the next speaker is Mister Ricardo Carlos from Portugal.
Thank.
Ricardo CARVALHO
You, madam chair. Well, colleagues, the impact of AI on electoral processes isn't just a technological issue. First and foremost, it's a democratic issue, and this is key for us in Europe. The use of AI systems enables highly sophisticated content to be created with a level of realism, which makes it very difficult to distinguish between what's true and what's false. And this clearly can produce manipulation of public opinion as well as disinformation.
I mean, this is automated disinformation, which is personalized and produced on a vast scale. So this is a massive danger, and we really need to pay attention to preventing it. This obviously has direct implications on the integrity of the electoral processes. Now the EU has the Digital Services Act and the AI Act. It's a pioneer in this kind of regulation, but that's not enough when it comes to protecting our democracies.
We believe that possibly it would be necessary to go further. We're talking about future, re possibly going to need future treaty reform at some point. Now, in this electoral context, there are three main challenges here. Firstly, transparency. Who is publishing this digital content, and how is it being generated?
Is it being generated by AI or not? Then, responsibility. How do we check who is creating this content and who is responsible for it if it's AI content? And then equality of access is the third issue. I mean, obviously, those who have more financial resources and more technological resources could have an advantage.
So there are many challenges that we face, which aren't only technical in nature, they're also constitutional issues that we need to look at. Thank you.
Gabriele Bischoff
Thank you very much. And, followed by MEP, Mister Jimenez Larras, also from Portugal.
Borja Giménez Larraz
Yes. Thank you very much, sir. Their colleagues, electoral integrity is under unprecedented pressure. AI-generated disinformation. The fake uncoordinated bot networks are no longer a future threat.
They are actively shaping campaigns across Europe right now. Russia has turned AI-driven disinformation into a strategic weapon to attack the legitimacy of the European Union, to divide our societies. We have seen it in the Romanian elections, in the French election, during the separatist process in Spain, also during Brexit. Let me say something. The fragmented structure of our electoral system makes us more vulnerable, as the AFCO stock taking briefing on the 2024 election shows.
27 different national frameworks for the European election create 27 different points of entry for malicious actors. We need to address this as a matter of urgency. And yet, AI is also part of the solution. It can detect coordinated and inauthentic behavior, flag foreign interference in real time, and help citizens access verified political information at scale. The question before this parliament is one of institutional courage.
Do our legal frameworks match the speed of this threat? Thank you.
Gabriele Bischoff
Thank you very much. And the next speaker is Miss Elizabeth Elisabetta Gardini.
Elisabetta Gardini
Thank you, madam chair. Well, I want to focus on this year. I think it really is key, because the idea of electoral integrity, which is what Mr. Caffallo Janis has dealt with in his report, is key. And obviously, we can't use yesterday's tools to tackle it.
With some Italian colleagues, I'm working to test some AI apps by asking them questions to which we know the answers. And these are questions that are a little bit divisive ideologically. And very often, we find ourselves with apps providing answers that are completely incorrect. So, I mean, I don't know because I'm not a technical person. So I don't know how we can regulate and govern this, because we're talking algorithms.
These are and the algorithms are fed by something. Now I know about history, so I continue to ask questions. And, eventually, I get the right answer. But it's only because I carry on answering questions that I get the right answer. I don't stop at the first response.
But a young person who doesn't know the information that they're looking for, well, they might stop at the first response. So that really is a key point. And that's why I think that mister Keflagiannis' proposals are something that we should take very seriously. And above all, we need to speed up here because the mood of the world, excuse me, is developing very rapidly, and we're struggling to keep up. Thank you.
Andre Diallao
Thank you very much, Missus Chair. Artificial intelligence is both an opportunity and a serious risk for democracy. We should not allow the AI to take over roles that belong exclusively to human beings: administration, justice, education, politics. AI has no moral conscience, no sense of responsibility, and cannot replace human judgment. It is deeply concerning that the European Commission is turning artificial intelligence sometimes into an instrument of control over public discourse, as well as using the digital service act to censor politically incorrect opinions.
We see how the commission is forcing platforms to use AI and the algorithms for proactive moderation and preemptive censorship in order to detect and eliminate so called systemic risks, disinformation, negative effects on electoral processes. These concepts are so vague that they lead to the automatic removal of legitimate political content. This results in overcensorship, legitimate political speech, satire, and uncomfortable opinions are suppressed automatically to avoid big fines. Thus, the AI becomes a tool of totalitarian control. In 2025, the commission fined platform x €120000000, the first major penalty under the DSA, using as a pretext the lack of algorithmic transparency and the claim that the platform had not censored enough disinformation and hate speech.
This is not just regulation, nor is it the protection of democracy or of the citizens. It is a new form of digital repression. The Romanian case is emblematic, the cancellation of the 2024 presidential elections. In Romania, TikTok's algorithms were invoked as the main pretext for annulling elections. For the first time in the history of the European Union, the constitutional court annulled the first round of presidential elections, citing massive manipulation by TikTok and other platforms, artificial amplification of content by algorithms.
Gabriele Bischoff
Thank you very much. Sorry. I already indicated, because we still have quite a few speakers on our list.
Missus chair.
Gabriele Bischoff
And want to give everyone the opportunity, and also around the same time.
Exactly. Thank you. Speech.
Gabriele Bischoff
Thank you very much. And I have Miss Carolina Marques from Portugal.
Thank.
Carolina Marques
You, madam chair. Colleagues, well, AI often focuses on the issue of disinformation and producing it. But AI also has positive potential for accessing and producing information respecting our values, as we've already heard today from one of our speakers. But there is a structural risk that we need to reflect on. We're seeing decision making constantly more and more effectively affected by this idea that we need instant decisions.
Speaker 63
And we're seeing a generation that's more and more used to scrolling and taking instant decisions, and is also influenced by populism. So what we need is something that provides high value information, because otherwise, everyone's getting anxious as a result of all this scrolling and this instant consumption. Social networks are molded by algorithms, and they try to generate reactions and polemics. What we need to do is try and stimulate people's thought, because at the moment, that's not happening with social media. AI doesn't only relate to content.
Carolina Marques
It also relates to how we have an influence on how we take political judgments and also how we relate to content. I mean, whether it's true or false, for example. So we need to include the fact that the way that people think is changing in electoral laws. Are we looking at that, or are we just seeing it as a secondary problem, which I wouldn't say it is? Thank you.
Adam Gumola
Thank you, madam president. You're getting better and better on Sword Dimes, so congrats. And distinguished guests, dear MP colleagues, as we discussed the rapid advancement of artificial intelligence, we must be honest. Our democratic institutions are currently outpaced by the technology meant to service, and there is one strong example for that. I mean, the political personalized advertisements.
If we couldn't control it, then we practically ban it. And this was the need of the time. But this is just one of the tools how, enemy has an interference, in our, in our elections. And I want to say that AI is no longer just one of these tools for productivity. It has become a weapon of mass distortion.
We have seen this clearly in the recent elections among our sisters and brothers in Romania, the digital interference orchestrated by Russia was not just a collection of fake news. It was a sophisticated AI driven operation designed to hijack the public discourse, exploit the social divisions, and manipulate the very emotions of the electorate. By using deepfakes and bot networks, they didn't just change what people thought. They changed how people felt. In Poland, being a part of the eastern border of the European Union, we feel the same that we are at the very front line of this battle.
And I want to say, and I want you to hear me out loudly, that there will be no other history like in Romania. There will be no other history when one member state left solely in this battle with the pressure of disintegration for the European Union from the East, but also from the West. We win. So we need to take very specific actions we discussed today, and please be very specific on that. Thank you very much.
Thank you. Thank.
Antonino Iaria
you. Well, political disinformation that undermines democracy has a long history. I mean, the press used to influence votes in the past, then we had television, radio, and.
Antonino Iaria
You have AI. Unlike the previous two, obviously, this is mass-produced disinformation. But if we're gonna try and fight it by judging what's true and what's false automatically, I think we cause problems. So the issue is the same as it always was. I mean, what did we do for fake TV information and fake newspaper information?
I mean, what we had was a counterpoint, which was true. So what we have to try and do is push the true facts. And this goes back to what I said earlier, which relates to education and an understanding of how AI works, because otherwise, we'll constantly have laws in place that block AI, and this will create a distortion. I mean, the fact is AI can also provide accurate information. So let's move forward massively with literacy, computer literacy, and that'll make it possible for people to decide what's true and what's false, and build opinions appropriately, which will protect democracy.
Milan Radin
Thank you, chair. Thank you very much. I will not use too much of your time. I just have to note how different the tone of this second panel is from the tone of the first panel. The difference in opinions on the same subject is quite noticeable.
I have to agree with most of what has been said in this second panel. However, let's face it. AI is not new print or new TV or the new Internet for the elections. I think that we can think of it as maybe a revolution between black and white TV and widespread color TV in houses of everyone. That being said, I do think that AI already influences elections, and it will continue to influence elections disregarding some total ban, which I don't think is plausible.
Serbia is a candidate country. We will fully implement whatever this esteemed parliament thinks would be the best, because we also quite understand the dangers of the AI. However, please let it be noted that there is a thin line, a very thin line, between regulating something which I perfectly agree on needs to be regulated and stifling what would be a new industrial revolution, which in the worst case scenario can lead Europe to lag behind the others. The others who will definitely either try to use AI to deploy it against Europe or who will just try to see where they can go with it. That being said, again, this is also a question whether the freedom of expression will be touched upon when the regulation of the AI and the Internet will be implemented.
Those are the problems to which I don't have a solution, to which I can't really give any meaningful input, but they are the problems of our age, and they are the problems with which we will have to live with the solutions as well. Thank you very much.
Gabriele Bischoff
Thank you very much. And, I have Miss Patrizia Gonzalez from Portugal.
Speaker 43
Well, this debate on AI and democracy needs to focus on electoral integrity and disinformation, but that's not sufficient, because the impacts of AI are more structural and deep-rooted in nature. They transform the way we get information, and they change the way in which public information is produced. We need to protect the functioning of democracy, not just elections, and this requires a structural response to boost digital literacy. Now this response isn't starting from nothing, from scratch. It's recognized at the European level.
It's recognized that education is key when it comes to social inclusion, access to information, etc. 80% of adults should have basic digital literacy. That's the goal. But we're far from this goal at the moment. A significant share of the population continues to lack the necessary tools to navigate an increasingly complex digital environment.
So when we talk about digital literacy, we're talking about clear policies, integrating this into educational policies, training teachers, investing in lifelong learning, and ensuring fair access to training resources. But there's also something that is even more important, which is providing education to citizens so they can understand how algorithms produce information, how information is filtered, and how AI algorithms can influence the behavior and decision making. This is all crucial because without this, any kind of regulatory activity will be insufficient. So the response is clear. We need to invest in our citizens' futures when it comes to AI, rather than reacting just to the symptoms of what's happening.
Speaker 68
We need to invest in rules, yes, but we also need to invest in people's ability to understand what's going on. Thank you.
Oviedev Cicimperan
Thank you very much. So I really appreciate the previous interventions. I agree that this information, fake accounts, and coordinated online manipulation are increasingly used to undermine trust in democracy. But what we can do as a concrete example, on many platforms, identity verification is a paid service now. The blue check can be bought by subscription.
Authenticity has become a product. Citizens should not have to pay to prove who they are, while fake accounts continue to spread manipulation. We think that we need a European framework for identity verification on social media. So this blue check should be for free, and this will reduce fake accounts, limit hostile influence, and increase transparency, while fully respecting fundamental rights. Thank you.
Gabriele Bischoff
Thank you very much, and thank you really for this debate. I think it shows really that we are in the middle of a huge change where we don't know how the future will be, but we are trying to set everything right, to address it and also to be politically able to influence it. And I think I would like to summarize four points. I think what was very clear also from UN under secretary general, mister Singh Jill, this sounds very simple, but it's very complex to say it all. You can only govern what you know.
And I think this is very essential for the work we want to do. We are also elected to do as parliamentarians, also to ensure that we have free, fair, and resilient political processes, and also in the age of AI. And at the same time, also, that we have to see when we look at the risks of, especially when we look at influence on elections or key decisions. Brexit was also mentioned here. This is not the future.
This is the reality we have to face here, and to see that we are well equipped to address it. And the problem also that we have is the question of speed, because these systems are developing so fast. And how can we make sure as parliaments that we don't only react to them, but that we actively take part in addressing it and shaping it from the beginning? And this also was mentioned also in the introduction, whatever regulation you do, it has to be very smart. And, also, I mean, that was highlighted here.
The balance is really essential between, on one hand, regulating and change that is so fast so that you don't just, we have this picture in Germany, of the rabbit and the hedgehog that you are always running, and AI is already a step ahead of you and how we address it, and especially, how we address on European level as national level this question, how can we guarantee this, really, that elections are indeed not influenced by AI in a way that undermines really what we want as free, fair, elections also and electoral processes, but also thank you for highlighting the opportunities that AI gives us to renew a democracy here and to use it for a democratic renewal in terms of participation, in terms also of more transparency that is possible, but also accountability and holding politicians accountable. These are the challenges we have to address here, and we will continue this structured dialogue, focused, and maybe also then focus deeper on one aspect in the future and discussing it. I think that might be a very good way how to proceed. I thank everyone here who participated in the debate. I think it was a very good and lively exchange, especially my thanks to the members of the national parliaments for their presence and really active participation.
And my job is also here to introduce you how you find your way after I declare this AFCO ICM meeting closed, to the reception and the dinner we will be holding in this building. And it will take place on the 3rd Floor in this building, the 2nd Floor. And you just go two levels down here, so you don't get lost. It's in the same building. And we are looking forward to meet you all for the dinner, and where we can. It's
Gabriele Bischoff
Yes. The now. Dinner. We are in time here to have you all there, and that maybe some informal exchanges can continue during this dinner. I also have to announce that the next meeting of the AFCO will be an extraordinary one that is very important for our members here of the European Parliament. It is scheduled to take place in Strasbourg on 04/27/2026 from 07:00 in the evening until 08:30 in the evening, in the Madriaka Building, and you will be informed.
And the next coordinator's meeting is scheduled for 05/05/2026 in Brussels. So thank you very much. Big thank you and applause for the interpreters. They made it possible that we can communicate in real life, and also we value very much the quality that we have here with real human beings doing the job. And with this, I see you hopefully at the dinner or at one of our next meetings.
Thank you very much, and thank you to the secretariat for organizing everything so perfectly. Thank you.
Ask anything about this video — I analyse its transcript to answer.