Meta rebuffs Brussels over AI rules

Meta just blew a hole in a European Union plan to tame artificial intelligence models.

The technology giant on Friday was the first Big Tech company to come out saying it will not sign the EU’s code of practice for general-purpose AI.

In a comment on Friday, Meta’s chief global affairs officer Joel Kaplan said the code “introduces a number of legal uncertainties for model developers as well as measures which go far beyond the scope of the AI Act.”

The aggressively lobbied code of practice was released last week and is the latest step by the European Commission to limit risks posed by AI models like OpenAI’s ChatGPT or X’s Grok. It comes as Grok is under fire for spewing Hitler-praising comments and other harmful responses.

At its core, the EU’s code of practice is an attempt by officials to get AI firms to follow the bloc’s rules without them having to launch full-fledged investigations.

It is designed to instruct companies on how to comply with the bloc’s Artificial Intelligence Act, a binding EU law. Companies that decide not to sign up face closer scrutiny from the Commission in the enforcement of the AI Act.

But the code has faced months of fierce lobbying from the tech industry.

Kaplan on Friday brought up a letter signed by over 40 top European companies in early July, including Bosch and SAP, which called on the Commission to pause the implementation of the AI Act.

Meta “shares concerns raised by these businesses that this overreach will throttle the development and deployment of frontier AI models in Europe and stunt European companies looking to build businesses on top of them,” Kaplan said.

Google in February also issued a stark warning that the bloc’s AI law risks hurting European innovation and growth, calling the code of practice a “step in the wrong direction.”

The Commission meanwhile has defended the initiative throughout its drafting.

Companies that sign up for the rules will “benefit from more legal certainty and reduced administrative burden,” Commission spokesperson Thomas Regnier said when the code was presented last week.

In response to Meta’s announcement on Friday, Regnier said in a statement that the code is a “voluntary tool, but a solid benchmark” and companies that chose to “comply via other means may be exposed to more regulatory scrutiny by the AI Office.”

French AI company Mistral on Thursday became the first to announce it would sign on the dotted line. OpenAI has also pledged to sign up.

This article has been updated to include a comment from the European Commission.