After three days of heated debate, on December 8, 2023, the European Parliament and the Council finally reached a provisional agreement on the “Proposal for a Regulation laying down harmonized rules on artificial intelligence” (the “AI Act”).[1]
The AI Act, along with the AI Liability Directive and the Revised Product Liability Directive[2], is integral to a broader strategy aiming to bolsteruser protection, particularly in the realm of high-risk AI systems. For instance, in the context of litigation, this new legal framework will, under a substantial perspective, prohibit or limit risk-related activities and, under a procedural perspective, simplify the burden of proof carried by claimants and provide efficient tools to obtain disclosure of evidence from AI providers. Anticipated consequences include a likely increment of litigations.
As far as the AI Act is concerned, its final text is not yet available, but it will be published in the near future. Then, the AI Act will enter into force after two years.
We will follow up with a more in-depth analysis as soon as the text becomes available.
In a nutshell, the agreement maintains the risk-based approach adopted in the previous versions of the proposal, but it introduces new rules for high-impact general-purpose AI models and high-risk AI systems, both subject to stringent transparency requirements, as well as new prohibitions on certain practices.
To strengthen protection of individuals, a fundamental rights impact assessment will need to be conducted prior to placing a high-risk AI system in the market. Moreover, the AI Act provides an obligation to indicate whether content is AI-created and limits the use of copyrighted content in training algorithms.
Real-time remote biometric identification systems represent one of the most significant changes. These may be used in public spaces only when strictly necessary for law enforcement purposes—to find victims of specific crimes, to prevent present or foreseeable threats (e.g. terrorist attacks), and to search for people suspected of serious crimes.
The provisional agreement also introduces a new governance system, including the creation of an AI office in Brussels. The office will enjoy a certain degree of autonomy and will be granted enforcement powers in tandem with the relevant national authorities.
