While Member States were at work finalising their adoption of the AI Act, policy and industry leaders discussed the near and medium-term future of AI in the EU
EU leaders gathered to discuss impending AI Act
On 30 January 2024, the Financial Times and Google hosted an event on ‘A Collaborative Approach to Artificial Intelligence (AI) Implementation and Regulation’. It featured panels with AI industry leaders, EU regulators and policymakers, and civil society experts all considering how the EU’s AI Act will impact both the safety of the technology for European users and the global competitiveness of European AI firms. Discussion also included reference to the Council of Europe’s forthcoming Framework Convention on Artificial Intelligence, Human Rights, the Rule of Law and Democracy, which would be the first binding international treaty on AI governance, following on from voluntary work underway through the G7 Hiroshima Process on AI as well as the Global Partnership on AI.
Industry cites an EU strength in AI deployment but notes regulatory roadblocks
In a panel billed as ‘the industry view’ on making AI a success in the EU, developers and deployers of AI noted that the bloc was well-positioned to emerge as a global leader in innovative use cases for AI solutions across various sectors. Speaking broadly for the tech sector, Cecilia Bonefeld-Dahl (Director General, Digital Europe) noted that the discourse on AI in the EU must not only acknowledge the right to be protected from potential harms of the technology but also the right to benefit from AI applications in healthcare or energy. Similarly, Jonas Andrulis (Founder and CEO, Aleph Alpha) noted that the EU never experienced a period of fallout from dangerously unregulated AI, suggesting that the AI Act was a solution seeking a problem to some extent. As a central EU challenger to US-based dominance in large language model (LLM) development, Andrulis also noted that the EU risks squandering its strengths in AI adoption by instead dedicating limited talent pools to compliance functions instead of innovation. Bonefield-Dahl offered perhaps the most tangible criticism of the pending AI Act for EU challengers in the sector, noting that it will cost a 50-person company approximately €300k (£256k) to implement and comply with the AI Act.
Procedural updates and clarifications on AI Act implementation
EU policymakers and regulators also joined to provide updates on the progression of the AI Act through the legislative process as well as information on the expected implementation of the law. Co-rapporteurs Dragoș Tudorache (MEP) and Brando Benifei (MEP) anticipated a parliamentary approval of the text by the end of March 2024, following on from the approval by Member States on 2 February 2024. Benifei was quick to suggest however that voluntary early compliance with some AI Act provisions could lead to greater public safety more immediately, especially in the context of the looming 2024 elections in the EU. In addition to enforcing the safety and security provisions of the law, Roberto Viola (Director General, DG Connect) took time to describe the responsibilities the new EU AI Office will have to support innovation. Viola referred to the mandate of the office as a “dual ecosystem” for innovation and trust, as the new institution will play a role in administering an expected €4bn (£3.4bn) in funding for AI Factories and the GenAI4EU promotion initiative among other innovation measures. Despite championing the successes of the AI Act, the MEPs did also mention the work remaining to address the existential threats posed by AI which they suggested could only be handled at a global scale. While the regulatory agenda on AI in the EU appears stable at the moment, the project of implementing the law amidst the rapid pace of technological development will keep European policy and industry leaders busy for the foreseeable future.
