MEPs debate on the world’s first set of comprehensive rules to manage AI risk and promote uses of AI that align with EU values including human oversight, privacy, and non-discrimination. The rules follow a risk-based approach and establish obligations for providers and users depending on the level of risk the AI can generate. They would prohibit AI systems with unacceptable levels of risk. They are also proposing tailor-made rules for generative AI, like ChatGPT, including labelling AI-generated content and making publicly-available summaries of copyrighted data used for training. Opening statements by Brando BENIFEI and by Dragoş TUDORACHE, Rapporteurs, by Eva MAYDELL; Marcel KOLAJA; Axel VOSS by Margrethe VESTAGER, Thierry BRETON, and by Susana SOLÍS PÉREZ and Josianne CUTAJAR. The transcript is the result of an automated process. Please check against delivery.

statement

Strasbourg

plenary session

VOSS, Axel (EPP, DE)

use of robotics and artificial intelligence (T)

VESTAGER, Margrethe (DK, CE)

BENIFEI, Brando (S&D, IT)

artificial intelligence

MAYDELL, Eva (EPP, BG)

KOLAJA, Marcel (Greens/EFA, CZ)

TUDORACHE, Dragoş (Renew, RO)

BRETON, Thierry (FR, EC)

Artificial Intelligence Act: opening statements by Brando BENIFEI and by Dragoş TUDORACHE, Rapporteurs, by Eva MAYDELL; Marcel KOLAJA; Axel VOSS by Margrethe VESTAGER, Thierry BRETON, and by Susana SOLÍS PÉREZ and Josianne CUTAJAR

https://api.multimedia.europarl.europa.eu/documents/20125/70732350/1686232303635_I242314Chapters%5BSD-EN%5D.srt

SOUNDBITE (Italian), Brando BENIFEI (S&amp;D, IT), Rapporteur, Thank you, President. The whole world is watching us. While alarming signals about A.I., the threat it said to pose to human race. Well these alarm signals often come out from the same big tech companies who are apprehensive about what they've actually created themselves.

SOUNDBITE (Italian), Brando BENIFEI (S&D, IT), Rapporteur, In the meantime, Europe has made headway, though it hasn't ceased in its efforts in recent months to come up with a palpable response to the risks which already exist, not just those that may come on in the future. Powerful technologies such as artificial intelligence, they say Europe will never be competitive without this.

SOUNDBITE (Italian), Brando BENIFEI (S&D, IT), Rapporteur, But some of the firms are agreeing this now, we do need rules. So I'm asking Parliament to provide us with a mandate for negotiations on the first piece of legislation in the world on this. Our position contains many improvements on the Commission's proposal, clarifying the text and bringing it up to date.

SOUNDBITE (Italian), Brando BENIFEI (S&D, IT), Rapporteur, Legal certainty, avoiding overlap with existing legislation, stimulating innovation, but at the same time keeping human beings and their fundamental rights at the heart of all our endeavours. That must remain the centre point of Europe's approach to artificial intelligence.

SOUNDBITE (Italian), Brando BENIFEI (S&D, IT), Rapporteur, Just to quote a few of the key points in our report, we have a clear definition and in line ith the work of international organisations with a chapter on Chat GPT that was originally admitted to avoid risks of discrimination and abuse of illegal content, then watermarking so that you can identify AI generated products.

SOUNDBITE (Italian), Brando BENIFEI (S&D, IT), Rapporteur, Article 52 provides this, but we we need to spel it out more clearly in the trialogue. We haven't overlooked either the risk to the environment. If you have algorithms with millions of pieces of data. Public authorities using artificial high risk artificial intelligence systems must have an impact study on fundamental rights so that there'll be a subsequent safeguard for vulnerable groups.

SOUNDBITE (Italian), Brando BENIFEI (S&D, IT), Rapporteur, We need to make sure there are judicial remedies available and proper explanation with collective responsibility if there's a negative impact on health and fundamental rights and adding reference to environment, democracy and the rule of law amongst the goals to be protected. Above and beyond the need to discuss the implementation of A.I.s and the impact on jobs that has to be done in conjunction with trade union representatives.

SOUNDBITE (Italian), Brando BENIFEI (S&D, IT), Rapporteur, Preventive policing is regarded as unacceptable. Biometrics. Indiscriminate use of web streams, in opposition to the use of biometric data in public spaces for surveillance purposes. The use of CCTV hasn't shown any palpable improvements in safety systems, but it does increase the risk of abuse and discrimination, particularly towards certain people.

SOUNDBITE (Italian), Brando BENIFEI (S&D, IT), Rapporteur, Women, non-white people and children. The risks to our countries and our democratic intentions. So we'll be fighting in the months to come to defend this approach, squaring up to risks to our freedoms. Some of the internal affairs decisioning, including a part of Italy, will be facing us. Now that said, there are a lot of positive prospects for increasing productivity and the quality of daily life.

SOUNDBITE (Italian), Brando BENIFEI (S&D, IT), Rapporteur, So we need to compare notes with lawmakers all around the world. We need a common approach. So the high impact and the code of conduct for generative A.I. have to be part of our work.

SOUNDBITE (Italian), Brando BENIFEI (S&D, IT), Rapporteur, Provided we don't replace the binding rules we pass if large efforts or governance to weaken these efforts, well, we need to make sure we strengthen our democracies in the context of environmental transition. We're going to do our very best for European citizens together with the Commission and the Council. Thank you.

SOUNDBITE (English), Dragoș Tudorache (Renew, RO), Rapporteur, Dear commisioner,   dear colleagues, as we move into this debate, I want to address the one question that I got the most over the past four years working with many of you in the committee and then working on this file: is this the right time? Is this the right time for Europe to regulate?

SOUNDBITE (English), Dragoș Tudorache (Renew, RO), Rapporteur, My answer is resolutely yes, it is the right time because of the profound impact the AI has. And I am convinced that all of you today will be bringing these arguments to the fore. It is the right also because hoping that companies will self-regulate is not enough to safeguard our citizens and our societies.

SOUNDBITE (English), Dragoș Tudorache (Renew, RO), Rapporteur, And we are now hearing more and more of the digital giants out there fearing their own products. It is also the right time because we've done preparations and worked seriously, responsibly on this file. And I want to start by thanking the commission for having the words, the foresight, already some years ago to start preparing these rules and put them forward.

SOUNDBITE (English), Dragoș Tudorache (Renew, RO), Rapporteur, I want to thank the leadership of all political groups and the leadership of this House for having the foresight to establish a special committee on artificial intelligence because we are now better prepared as lawmakers and I think we've ever been to actually tackle these rules. And now let me move to the rules themselves.

SOUNDBITE (English), Dragoș Tudorache (Renew, RO), Rapporteur, What will we vote tomorrow? I'll walk you through some of the key elements, and many thanks to my co-rapporteur, Brando Benifei, with whom I worked very well, for already pointing some out. I will try not to repeat them. We have worked hard to provide a definition of AI that is not only agreed by all political groups in this House, but also aligned with that of the OCD and other partners such as the US.

SOUNDBITE (English), Dragoș Tudorache (Renew, RO), Rapporteur, This gives the democratic world a starting point for developing a common terminology. When discussing it. We are providing, throughout text developed in Parliament, start-ups and SMEs in Europe with a strong network of sandboxes so they can innovate, they can grow and develop technology in accordance with European values and in respect to fundamental rights. Increasing the trust in their products as well as their competitiveness.

SOUNDBITE (English), Dragoș Tudorache (Renew, RO), Rapporteur, We have worked to put in place a coherent governance structure that makes sense. It is meant to ensure that we consolidate the digital single market, that member States have the proper tools and support in place for the enforcement of this regulation, that Europe has a futureproof approach to new developments in AI and that we don't repeat some of the mistakes we've done with GDP over ten or even five years down the line.

SOUNDBITE (English), Dragoș Tudorache (Renew, RO), Rapporteur, It is this governance structure that will give Europe the ability to deal with the rapid evolution of the AI and to reap the most benefits from it. And we have worked first and foremost to ensure our citizens rights and freedoms are not just respected but protected and strengthened. We don't want mass surveillance, we don't want social scoring. We don't want predictive policing in the European Union. Full stop. That's what China does, not us.

SOUNDBITE (English), Dragoș Tudorache (Renew, RO), Rapporteur, We want all public authorities, whether they distribute benefits or whether they force the law, to be accountable, transparent, focused on our rights with strict democratic oversight. Last but not least, I am proud that we had the foresight to devise a preliminary set of rules for the AI took many by surprise and by storm that is generative AI based on powerful foundation models.

SOUNDBITE (English), Dragoș Tudorache (Renew, RO), Rapporteur, We worked on rules that allowed technology to flourish and develop while ensuring that the models, potentially used by millions of smaller companies or start-ups, as well as by hundreds of millions of individual users that they are safe.

SOUNDBITE (English), Dragoș Tudorache (Renew, RO), Rapporteur, We introduced obligations of diligence for content generated by these models and of transparency regarding copyrighted material using their training and very importantly, we found overwhelming consensus among all political groups on these rules. And that, I think gives a very powerful statement.

SOUNDBITE (English), Dragoș Tudorache (Renew, RO), Rapporteur, Colleagues, I take full responsibility for my words. We are making history, we are writing history. We are shaping the global development of AI in the right direction with humankind and human interest at the very centre. Thank you very much. And I really want to give a special thanks to all the shadow rapporteurs for how we have come together around this text that we vote tomorrow. Thank you.

SOUNDBITE (English), Eva MAYDELL (EPP, BG), on behalf of ITRE, Regulating artificial intelligence may at first sound like a very technical matter, but it is so much more than that. It is about asking ourselves the big question, what kind of society do we actually want to live in? And to me, the answer is very clear. A society in which A.I. enables our democracy and not one in which it undermines it.

SOUNDBITE (English), Eva MAYDELL (EPP, BG), on behalf of ITRE, But of course, this is much more easier said than it is actually done in practice. For us to get there I think we should not be playing Whac-A-Mole or regulating the next big thing. It's about putting in place effective guardrails. It's about putting principles and mechanism that will last for at least the next couple of years to come. But it is also about working with our partners.

SOUNDBITE (English), Eva MAYDELL (EPP, BG), on behalf of ITRE, It is about working with international partners to shape a global governance framework. It's about maximising the potential of AI, while in the same time addressing the existing and looming concerns. So, no, it is not technical. It Is about the future of our democracy. Thank you.

SOUNDBITE (English), Marcel KOLAJA (Greens/EFA, CZ), on behalf of CULT, Madam Executive Vice President, Commissioner. Colleagues, today we are rewriting history. The next era will be the era of artificial intelligence. And what we are about to do in this House is to give guarantees, guarantees that European citizens will benefit from the development of AI, that they will not be discriminated. That AI will not be abused to spy on them.

SOUNDBITE (English), Eva MAYDELL (EPP, BG), on behalf of ITRE, We need to ensure that malfunctioning technologies will not ruin students future careers. That's why. Technologies to proctor students during exams will be categorised as high risk. Let me also speak about spying on citizens. We do not want governments to be able to identify everyone in every moment. To records where we walk, which places we visit, or who we talk to in the streets.

SOUNDBITE (English), Eva MAYDELL (EPP, BG), on behalf of ITRE, Imagine your favourite example of the worst politician in your country. Now, would you want them to have a remote control to such spying? I do not. I believe that the use of facial recognition in public spaces must be banned. Thank you.

SOUNDBITE (GERMAN), Axel VOSS (EPP, DE), Thank you, President. Madam Vice President. Commissioner. Colleagues, I would like to start by thanking everybody for the excellent cooperation which has enabled us to reach some areas of compromise. The importance of AI is, something which we should start by reiterating, this will be the basis of all future developments. This is why it's so important that we strike the right balance. We have to build on the 2020 White paper so as to ensure that we have an ecosystem for trust. This proposal, I think, has retained the ecosystem of trust.

SOUNDBITE (GERMAN), Axel VOSS (EPP, DE), At the same time, however, all this proposal is driven by fear, mean that this proposal is not balanced enough. And that, I think, will hinder efforts to become leaders in the world in terms of AI. So, in my view, the proposal we are voting, talking about today and voting on tomorrow should be tested so as to ensure that we can our industry turns this into a success story.

SOUNDBITE (GERMAN), Axel VOSS (EPP, DE), The Comission and standardisation organisations must now ensure that they act quickly so as to ensure they provide the support measures and establish technical standards. In my view, in the trial of negotiations, we should try to ensure that we eliminate over overlap and contributions. Without the sectoral rules and digital legislation, we have to ensure we eliminate all of those.

SOUNDBITE (GERMAN), Axel VOSS (EPP, DE), I think we should also ensure that within the European Union we do indeed ensure that we have the same rules everywhere. We should try to ensure that we don't repeat the mistakes of the GDPR. And I think it's important that we ensure that data sets be properly trained. I think there's real room for improvement there.

SOUNDBITE (GERMAN), Axel VOSS (EPP, DE), And one final point. Small and medium sized enterprises and start-ups will run the risk of paying the price with all these expensive rules on air. We have to ensure that they have access to waivers and perhaps to ensure that we can intervene in the supply chain so as to ensure that we can pave the way to a Europe wide success in this new field.

SOUNDBITE (English), Margrethe VESTAGER, Executive Vice-President of the EC in charge of Europe fit for the Digital Age, This is indeed big. It is historic. This debate and the vote tomorrow. I want to congratulate the European Parliament, the rapporteurs and the shadows for this impressive work, not only in relating to the proposal, but also in the preparation and the in-depth understanding of what it is that we are dealing with. And I think this is what has enabled you to strike a compromise on striking 3300 amendments and developing a comprehensive legislative approach on general purpose AI.

SOUNDBITE (English), Margrethe VESTAGER, Executive Vice-President of the EC in charge of Europe fit for the Digital Age, An impossible task for the unprepared, nor even an easy task for the well-prepared, as the Parliament had prepared. But just a few words about why we table this proposal in the first place. Because, when it comes to technology, the most important thing when you embrace it is that you can trust it. And only by embracing something can you control it. If you keep it at arm's length, very difficult to get a tight grip.

SOUNDBITE (English), Margrethe VESTAGER, Executive Vice-President of the EC in charge of Europe fit for the Digital Age, And what we can do here is to create trust, legal certainty, to enable AI to develop in a positive manner.T o serve our society, our ambitions, the sectors, the businesses, the industry. We can do amazing thing with artificial intelligence when we can trust it. When we started our journey, I frequently got the question "Why are you regulating A.I.? This will hamper innovation. Why not just let it go?".

SOUNDBITE (English), Margrethe VESTAGER, Executive Vice-President of the EC in charge of Europe fit for the Digital Age, But I think by now everyone realises that this is the important moment. Yes, this is the time. And also this has matured the outer world to see that the risk based approach, the focus on the use of technology that is the future proof one. I think this is also a chance to put a big line under the Brussels effects.

SOUNDBITE (English), Margrethe VESTAGER, Executive Vice-President of the EC in charge of Europe fit for the Digital Age, The work of this Parliament is not to be seen in isolation, is it to be seen with the ambition of democracy, not only catching up with technology, but shaping how we use technology in the best interest of the human being. I commend you and I look very much forward to the debate, the vote tomorrow and the process from there on. Thank you very much,

SOUNDBITE (French), Thierry Breton, Executive Vice-President of the EC in charge of Europe fit for the Digital Age, Honourable members, before going any further, I would like to compliment the rapporteurs and shadows for the various committees and all the members of this House for the huge amount of work they have done on the Artificial Intelligence Act.

SOUNDBITE (French), Thierry Breton, Executive Vice-President of the EC in charge of Europe fit for the Digital Age, This legislative proposal is part of our efforts to organise the information area within Europe, in our internal market, revolving around four main pillars that I briefly mentioned before with digital platforms.

SOUNDBITE (French), Thierry Breton, Executive Vice-President of the EC in charge of Europe fit for the Digital Age, The creation of a single data market, any of the data securing the digital area with cyber solidarity, and now the Artificial Intelligence Act to back that all up in 2021, we tabled a proposal in the Commission.

SOUNDBITE (French), Thierry Breton, Executive Vice-President of the EC in charge of Europe fit for the Digital Age, It has to be said that altogether we find ourselves grappling with a very complex issue: this is an intersection of a wide range of interests and principles here. The proper framework for applications, respect for of values is so important, as is safety and security. But I feel profoundly convinced that we are the continent with the democratic system best placed to govern this regulation in a transparent, balanced, responsible report in its way.

SOUNDBITE (French), Thierry Breton, Executive Vice-President of the EC in charge of Europe fit for the Digital Age, Undoubtedly, what we have here is also the possibility of gradually phasing in a new world order. But let us return to the act itself. We didn't wait until people called upon us to take action in the union. Europe has shown that to think ahead of the curve because since 2021 we've all been working together on this regulation.

SOUNDBITE (French), Thierry Breton, Executive Vice-President of the EC in charge of Europe fit for the Digital Age, This endeavour is what has put us in the position to be the first ecosystem in the world to be debating clear, notable and proportionate rules in the face of some of the possible implications of AI, while making the most of technological progress and innovation at the same time. What we are doing is not pressing the pause button as you sometimes hear people sayin.

SOUNDBITE (French), Thierry Breton, Executive Vice-President of the EC in charge of Europe fit for the Digital Age, On the country, we're trying to react in a politically responsible way to provide a clear, regulated, safe framework. And that's what we're all doing together in Europe Now. Regarding Parliament's report, now under debate, I would like to welcome the approach, which is risk based. The Commission had proposed not, of course, but all the rapporteurs have run with this and also note that are the the report includes a definition of hiding of high risk and governance, amongst other welcome innovations.

SOUNDBITE (French), Thierry Breton, Executive Vice-President of the EC in charge of Europe fit for the Digital Age, Now, if I could say something about general use and generative A.I., that's something that lies at the heart of many international debate today. It seems to me that a major task force in negotiations will be to ensure that EU legislation provides palpable responses to the challenges faced by generalised A.I., particularly generative systems such as chatbots.

SOUNDBITE (French), Thierry Breton, Executive Vice-President of the EC in charge of Europe fit for the Digital Age, For example, I note that European Parliament, the Council and the Commission share their understanding of the need to take action and to ensure transparency regarding generative A.I.. To be clear, we need to label anything that is A.I. generated images, tags, or by using watermarking. This is a central element if we are to build up the confidence of providers and consumers. There are other challenges, such as bias and disinformation.

SOUNDBITE (French), Thierry Breton, Executive Vice-President of the EC in charge of Europe fit for the Digital Age, There's a crucial time just for our society, particularly as we're coming into a major EU point cycle of elections and many member States will be affected by our proposals. And so I feel that we need to work together in parallel with negotiations on the AI act, as we know, this won't come into force in 2026.

SOUNDBITE (French), Thierry Breton, Executive Vice-President of the EC in charge of Europe fit for the Digital Age, So that right as of now, once this has been adopted following the trialogue, we can send out a clear message to all players, of today. That this is the meaning behind the artificial intelligence pact put forward a few weeks ago: the idea is that we should support firms in implementing the future roll out of rules on AI once this legislation has gone through, of course. That's to ensure that the transition can go through as easily, harmoniously and fast as possible.

 Of course, we're not trying to pre-empt the negotiations to come, but we want the major guidelines through as quickly as possible and we're entering commitments with all the the stakeholders and we have positive signs. Lastly, ma'am, make it very clear that of course will accept no interference whatsoever at this stage of our democratic process in the European Union on such a central issues as artificial intelligence from anyone international for private interests.

 I welcome the fact that the code and the decision making progress on this, and I trust that Parliament will adopt this position tomorrow, which will allow us to move without any delay into the trialogue. You can rely on, on myself, the Commission and all our staff to work together with the two branche untill the agreement is forthcoming.

SOUNDBITE (Spanish), Susana SOLÍS PÉREZ (Renew Europe, ES), on behalf of ENVI, you very Thank much. President Commissioner, we are talking here about historic legislation. Europe has always been a pioneer when it comes to A.I. this is technology which can bring with it great benefits. But which clearly also poses risks in terms of fundamental freedoms and disinformation. This is why I very much welcome the focus in this proposal.

SOUNDBITE (Spanish), Susana SOLÍS PÉREZ (Renew Europe, ES), on behalf of ENVI, It draws a distinction in technology in use. We don't condemn AI per say, but risky practices. It's very clear to us that certain aspects are unacceptable, anything which is discriminatory or harmful, and in that case they should be banned. We need to have safe, transparent rules with accountability so as to ensure that high risk use, such as health applications, environmental transport and justice be properly regulated.

SOUNDBITE (Spanish), Susana SOLÍS PÉREZ (Renew Europe, ES), on behalf of ENVI, We should also recognise that we need a flexible law which provides incentives for companies to use it so as to ensure that they can adapt quickly to the enormous progress which is being made. We've seen this recently with generative AI, Chat GPT in which time is pressing so as to ensure that we can establish rules.

SOUNDBITE (Spanish), Susana SOLÍS PÉREZ (Renew Europe, ES), on behalf of ENVI, Europe needs to take a big step forward and this law will help us establish trust amongst our citizens and ensure that we remain in the vanguard of efforts to develop AI.

SOUNDBITE (English), Josianne CUTAJAR (S&D, MT), on behalf of TRAN, Madame Commissioner, colleagues, the rapporteurs. It was a pleasure working with you from the transport side of things and today's vote places Europe at the forefront of becoming a global hub for cutting edge innovation, channelling the potential of fair, accessible, transparent and trustworthy AI for our citizens.

SOUNDBITE (English), Josianne CUTAJAR (S&D, MT), on behalf of TRAN, The Act strives to achieve an important balance encouraging investment, digitalisation and technological progress whilst regulating via a risk based approach, respecting our European values and fundamental rights.

SOUNDBITE (English), Josianne CUTAJAR (S&D, MT), on behalf of TRAN, Our strides in this field must continuously seek to improve our quality of life, relying on human centric models, keeping people in the driver's seat, while also ensuring that I serves as an enabler for inclusive mobility and transports. Well, this can only be achieved if we ensure that our workforce is sufficiently the skills and our SMEs are kept on board.

SOUNDBITE (English), Josianne CUTAJAR (S&D, MT), on behalf of TRAN, Fostering ethical AI, built on responsibility and clear liability will create a culture of trust for consumers and investors alike, incentivising the uptake of technologies that can improve the life of European citizens and the resilience of critical infrastructure, whether it is in the health and transport fields. Thank you.