EU AI Act: Trilogues begin about rules for safe and transparent AI in the EU

The European Parliament (EP) adopted at the Stratsbourg’s plenary session of June 14th, 2023, its negotiating position on the Artificial Intelligence (AI) Act with 499 votes in favour, 28 against and 93 abstentions. The adoption of EP’s position was preceded the previous day by a debate on the report from the leading joint committee on the matter – the Committee on the Internal Market and Consumer Protection (IMCO) and the Committee on Civil Liberties, Justice and Home Affairs (LIBE).

According to the debates, it should be stressed that the principle desire of the EP concerning the AI Act is to make the European Union (EU) a leader as regards regulation and innovation in AI as well as a body that takes into consideration open letters published since Open AI first released its groundbreaking AI tool. The debate’s keynotes therefore revolved around three principle themes.

The regulation should firstly be future-proofed in order to address the rapid evolution of AI. Therefore, issues such as AI-based disinformation (e.g., deepfakes), AI monitoring of employees in the workplace without their consent or even, automated decision-making in migration management should be taken into account.

Secondly, transparency and responsibility should be at the forefront of the EU’s regulation on AI. Generative AI systems, which are based on foundation models, like ChatGPT, would have to comply with transparency requirements (disclosing that the content was AI-generated, and also helping to distinguish so-called deepfake images from real ones). MEP Deirdre Clune, who represents the PPE Group, also emphasised the fact that developers of generative AI systems would have to be asked “how these systems were trained and how they were developed”. MEP Josianne Cutajar, rapporteur for the Committee on Transport and Tourism stated that “fostering ethical AI built on responsibility and clear liability will create a culture of trust for consumers and investors”.

The third theme focused on prohibited AI practices and in particular, the ban on the use of biometric identification systems in publicly accessible spaces. The debate at the EP showed that a certain degree of difficulty is still being experienced in terms of aligning MEPs on this issue. Certain MEPs were against the inclusion of exceptions to this ban, arguing that this would risk people’s rights being affected (e.g., mass surveillance) while others urged that AI be used for law enforcement (e.g., for terrorist attacks, criminal arrests, and to prevent child abductions); for certain deputies, a balance should be sought between ethical considerations and necessity.

It is worth remembering that the European Commission’s AI Act proposal outlawed facial recognition in public spaces, but included exceptions to the ban that cover situations involving missing children, dangerous criminals and terrorists. However, the EU’s key lawmakers working on the proposal have removed these exceptions and expanded the scope of the ban. The text that they propose would forbid the use of facial recognition tech in real-time and impose limits on the use of the technology that involves pre-recorded footage.

More than two years after the adoption of the AI Act proposal by the European Commission, the EP set the stage for the trilogues between the EU Institutions (the European Commission, the Council of the EU, and the EP) on the final shape of the law. The European Commission is pushing for the AI Act to be finished by the end of 2023. The first trilogue between the Council, the Parliament and the Commission began the evening after the EP adopted its negotiating position, which is indicative of the sense of urgency that prevailed at the time.

It is no coincidence that the above negotiating position has been adopted less than a year before the European elections of 2024 by a big majority at a time of intense debate on the impact of misinformation and external interventions by hybrid actors who, through the use of artificial intelligence, could expand the possibilities of manipulating democratic electoral processes. According to Euractiv’s Luca Bertuzzi, “negotiations will intensify once Spain takes over the rotating presidency of the Council in July, as Madrid has made finishing up the AI law its top digital priority”.

On a final note, it should be kept in mind that discussions are being held in the meantime at the Council of Europe (CoE) on the adoption of a convention on AI, human rights, democracy and the rule of law. The 6th meeting of the CAI Committee of the CoE, responsible for the file, took place in Strasbourg from 31 May to 2 June 2023. The European Commission’s mandate, which was received on 21 November 2022 from the Council of the EU (Council) to open the negotiations on behalf of the EU, risks being impacted in the forthcoming months by the trilogues on the AI Act at the EU level, as Member States may request that the text of the future convention of the CoE on AI aligns with the Council’s general approach to the AI Act proposal.

Like this article?
Share on Facebook
Share on Twitter
Share on Linkdin
Share by Email