According to an open letter signed by more than 160 executives at companies ranging from Renault (RENA.PA) to Meta (META.O), a new set of rules could jeopardize Europe’s competitiveness and technological sovereignty. The draft EU Artificial Intelligence legislation that European lawmakers agreed to this month will force systems like ChatGPT, which aims to create natural-sounding human conversations with robots, to disclose their AI-generated content, help distinguish so-called deep-fake images from real ones and ensure safeguards against illegal content. It’s part of an effort by the European Union to be a global leader in trustworthy and secure AI and to protect the values upon which its societies are built.
Based on the European Charter of Fundamental Rights, the EU’s draft rules aim to ensure AI can be used without violating its citizens’ privacy or civil liberties. Those include respect for freedom, diversity, pluralism, justice, equality, and the rule of law. The European Parliament is expected to vote on the legislation by the end of 2023.
However, the legislation’s scope needs to be narrower and would limit innovation in the fast-moving field of AI, experts say. For example, it focuses only on providers of high-risk AI systems rather than also on those that deploy them or those who use the systems for other purposes, they say. This could make it difficult to determine whether a system is, in fact, high-risk or not.
They are also concerned that the rules might impede international law enforcement cooperation. And they believe that the provision exempting law enforcement from these regulations will be abused to circumvent the legislation.
To address these concerns, European lawmakers have made several amendments to the legislation. A compromise text, which is due to be approved later this week, expands the scope of the regulation to cover those who develop and deploy AI systems and those who use them for other purposes, including those that may raise their risks.
It also requires providers of high-risk AI systems to prove they meet specific requirements before putting them on the market or in service. It would require them to describe and document those requirements while establishing a register of those systems across the EU.
Lastly, it would exclude from the scope of the legislation any systems designed to comply with existing EU law or used solely for legal and contractual purposes or in the exercise of public functions by entities established in the EU. This would ensure that the rules do not interfere with EU competition and internal market law and prevent divergences that hamper the free movement of AI systems across the bloc.
