Report from Reuters
In Brief – French President Emmanuel Macron’s address to a global AI Summit in Paris called for Europe to recalibrate AI regulation to support innovation and made the case that France, with its AI talent and nuclear-powered electricity generating capacity, was well-positioned to lead the region forward. Against a backdrop of US President Donald Trump’s deregulatory focus on AI, which included Vice President JD Vance telling the summit (video here) that EU tech and AI regulation was excessive, and that the US was committed to its AI leadership, the overall tone of the summit was far less about avoiding risks and far more about achieving benefits. Speaking of European AI regulation, Macron said, “It’s very clear we have to resynchronize with the rest of the world.” EU Commissioner Henna Virkkunen, who leads on digital policy for Commission President Ursula von der Leyen, also told the summit that EU will simplify its rules and implement them in a business-friendly way. In her address, von der Leyen said she intends to make the EU “the AI continent” and announced investments in “public supercomputers” to serve start-ups and scientists and the creation of 12 “artificial intelligence factories.”
Context – When the EU enacted its AI Act last year many forecast it would be the global model for AI regulation. We didn’t see it then and trends are in the opposite direction. President Macron was a voice of restraint during the AI Act talks, cautioning against regulating general “foundation models”, which includes the AI giants, but also European upstarts like France’s Mistral. The EU, pressed by the Parliament, went the other way. The first AI Act rules came into force on February 2nd, with the focus on prohibited edge-case AI uses. But under the radar was also the AI Act mandate requiring all providers and users of AI systems in Europe to ensure that all staff and other persons dealing with AI systems have a sufficient level of “AI literacy”, adding new training and compliance burdens to even small AI developers and low-risk system users. The AI Act compliance calendar is long and rules for the big general purpose AI models come out later this year.