EU strategy on AI: where are we at and what are the next steps



Maximising the benefits and the potential of digital and ensuring ‘a technology that works for people, a fair and competitive economy and an open, democratic and sustainable society’ are the objectives underlying the Union’s Digital Strategy defined in 2020. The strategy was later complemented by the Digital Compass – presented by the Commission a year ago, in March 2021 – which set further digital objectives to be met by the EU by 2030. Against this background, we can find many digital dossiers currently of interest for the European institutions. From the Digital Services Act (DSA), the ePrivacy Regulation, the revision of the Directive on Security of Network and Information Systems (NIS2) and the Digital Markets Act (DMA) for which a provisional agreement has just been reached after 4 rounds of inter-institutional negotiations. Adding to this, in the first few months of 2022 more initiatives have been set by the Commission to contribute to the above-mentioned goals and to forward the Union’s digital sovereignty and strategic autonomy. These include the Chips Act, the Data Act, and the Regulation on a space-based secure connectivity.

In order to take stock of the legislative process on Artificial Intelligence (AI) at European level, following, we will focus exclusively on this specific file, and report on its latest developments and the next steps.

Specifically, this week has been a defining one for this important file. On Tuesday 22 March, the Special Committee on Artificial Intelligence (AIDA) voted on its own-initiative report on AI in the digital age. The Special Committee was established by the Parliament in 2020 and has since then worked towards assessing the impact of AI on the EU economy in the digital age, focusing, among other matters, on skills, employment, fintech, education, health, environment, industry, energy, etc. The Committee was also tasked with the mission of defining a roadmap on ‘A Europe fit for the digital age’ for 2030. German MEP Axel Voss (EPP) was nominated Rapporteur for the adopted AIDA report. His report urged the EU to act, demonstrating that the bloc is currently lagging behind in the global digital competition race. When compared to China and the USA, the EU appears to be lacking in the ability to access and share high-quality data, harmonised rules and standards, funding, training and infrastructure for AI. The report also stressed the importance of AI features that aim to ensure the digital transition is human-centric and consistent with the Fundamental Rights and EU core values. The Parliament will be discussing and voting on the report during the Parliament plenary session in May.

Nevertheless, the AIDA report is meant to integrate a second legislative initiative within this domain, that is, the European Commission proposal for an EU regulatory framework on AI (AI Act), which the European Parliament and the Council are currently discussing. Indeed, a further important step for the European approach towards AI held this week was the joint meeting of the Committee on Internal Market and Consumer Protection (IMCO) and the Committee on Civil Liberties, Justice and Home Affairs (LIBE) which took place on Monday 21 March. During the meeting, MEPs had the chance to exchange views with experts on the issue at stake. Co-rapporteurs on the dossier are the Italian MEP Brando Benifei (S&D) for IMCO, and Romanian MEP Dragos Tudorache (Renew) for LIBE. It should be recalled that the trans-sectoral nature of AI applications gave rise to a dispute among committees over competences, following which, earlier this year, a third Rapporteur on the file for the Legal Affairs Committee (JURI) was appointed – the previously mentioned Mr. Voss. As a consequence, the discussion in the Parliament will be held under a joint committee procedure (IMCO, LIBE, JURI), while other committees have been assigned specific competences. The next joint IMCO/LIBE committee meeting is scheduled for April 2022, the two committees are expected to issue their reports by the end of September 2022 to be voted on in the plenary this autumn. Instead, the JURI Committee would like to be able to release its report by May so that it could still be considered by the other two leading committees for their respective reports.

According to the proposal, apart from being ‘excellent’, European AI should be ethical, trustworthy and beneficial to users and providers. If consensus has been reached around these guiding principles, other matters still spark off debate amongst the legislators. Namely, some MEPs argue the definition of AI is too broad and are concerned with the fact that low risk AI applications might also fall under said definition causing over-regulation. However, the Commission recalled the current definition had been drawn up by the AI High Level Expert Group, and is aligned with that provided by the OECD. Further elements of dispute regard self-certification and the link between conformity assessment standards (Art.43) and Annex 3, (listing high risk applications). According to the Article for high-risk uses listed in the annex, a self-assessment procedure would be sufficient to provide for the “CE” label. While some claim such certification should be provided by a third party, others argue it is sufficient for the provider to assess and self-certificate it. Different stances have emerged also with respect to AI uses where social scoring would be allowed only for a few limited exceptions authorised by a judge (e.g. for EU arrest warrants and imminent threats such as terrorism). Lastly, the risk of having algorithms and AI customisation processes which are biased has raised trust issues for AI developers, resulting in the quest for higher ethical standards for many.

In conclusion, despite the European AI Regulation being a world first, the EU must look to striking the right balance between ethics, innovation and safeguarding Europeans’ rights, while putting in place future-proof provisions. This effort has to be made in an attempt to meet its digital objectives, as well as to establish worldwide standards and become more attractive to the tech industry.