Meta refuses to sign EU’s AI code of practice

Meta has refused to sign the European Union practice code For its AI law, weeks before the rules of the block for providers of AI models for general purposes come into effect.

“Europe is going the wrong path on AI,” wrote Meta’s Chief Global Affairs Officer Joel Kaplan in one after On LinkedIn. “We have carefully assessed the practical code of the European Commission for General AI (GPAI) models and Meta will not sign it. This code introduces a number of legal uncertainties for model developers, as well as measures that go much further than the range of the AI Act.”

The EU’s practical code – a voluntary framework that was published earlier this month – is intended to help companies implement processes and systems to comply with the legislation of the block for regulating AI. The code requires, among other things, that companies provide documentation about their AI tools and services and are regularly updated and forbids developers by training AI on illegal content; Companies must also meet the requests of content owners not to use their works in their datasets.

The name of the EU’s implementation of the legislation “over -range”, Kaplan claimed that the law will “nip the development and deployment of frontier AI models in Europe and European companies will stunt that companies want to build on it.”

A risk -based regulation for applications of artificial intelligence, the AI ACT prohibits a number of “unacceptable risk” user cases, such as cognitive behavioral manipulation or social scoring. The rules also define a series of “risky” use, such as biometrics and face recognition, and in domains such as education and employment. The law also requires that developers register AI systems and meet the obligations of risk and quality management.

Technology companies from all over the world, including those in the forefront of the AI race such as Alphabet, Meta, Microsoft and Mistral AI, fought The rules that even encourage the European Commission to postpone its rollout. But the committee has held and says it will not change its timeline.

The EU also published on Friday guidelines For providers of AI models prior to rules that take effect on 2 August. These rules would find providers of “General AI models with systemic risk”, such as OpenAI, Anthropic, Google and Meta. Companies that have such models on the market before 2 August will have to comply with the legislation by 2 August 2027.