Meta declines to sign EU AI code of practice

Meta Platforms has decided not to motion the European Union’s (EU’s) codification of signifier designed to assistance companies adhere to the bloc’s AI Act.

This determination was announced by Meta’s main planetary affairs serviceman Joel Kaplan, successful a LinkedIn post.

“Europe is heading down the incorrect way connected AI. We person cautiously reviewed the European Commission’s Code of Practice for general-purpose AI (GPAI) models and Meta won’t beryllium signing it,” Kaplan said successful the post. “This Code introduces a fig of ineligible uncertainties for exemplary developers, arsenic good arsenic measures which spell acold beyond the scope of the AI Act.”

The EU’s voluntary codification of practice, formulated by 13 autarkic experts, aims to assistance companies successful aligning with the AI regulations.

It requires signatories to people summaries of grooming contented for their AI models and adhere to EU copyright laws. The codification besides mandates regular updates connected AI tools and services and prohibits the usage of pirated contented successful AI training.

Additionally, companies indispensable respect requests from contented owners not to usage their works successful datasets.

Meanwhile, different tech large Microsoft, is considering signing the code. Microsoft President Brad Smith told Reuters: “I deliberation it’s apt we volition sign. We request to work the documents.

“Our extremity is to find a mode to beryllium supportive and astatine the aforesaid clip 1 of the things we truly invited is the nonstop engagement by the AI Office with industry.”

Earlier successful July 2025, respective tech companies urged the European Commission to delay the AI enactment rollout, requesting a two-year postponement. However, the Commission did not alteration the timeline.

OpenAI and Mistral person already signed the code.

Recently, the European Commission besides released guidelines for AI exemplary providers. These regulations volition interaction providers of ‘general-purpose AI models with systemic risk,’ including OpenAI, Anthropic, Google, and Meta.

Companies with specified models indispensable adhere to the authorities by 2 August 2027.