Elon Musk’s AI company, xAI, has announced that it will sign the safety chapter of the EU’s Code of Practice for generative AI.
AI developers can voluntarily sign up to follow the Code in order to be considered compliant with the EU’s AI Act.
The Code is split into three chapters: transparency, copyright, and safety and security.
On Thursday, xAI announced that it will sign only the last of these three chapters – which lays out how companies should take steps to identify and mitigate safety and security risks that might arise via their models.
The company, which makes its Grok AI chatbot available through the also Musk-owned social media platform X, said that other parts of the AI Code contained requirements that are “profoundly detrimental to innovation”.
It specifically hit out at the copyright chapter, describing it as “clearly over-reach” – but without giving any further detail.
Signatories to the copyright chapter commit to only reproducing and extracting content under copyright if it’s lawfully available, among other provisions.
Earlier today Microsoft announced that it will sign the GPAI Code in full. However, social media giant Meta previously said that it will not sign.
Responding to a question about xAI at today’s midday press conference, a Commission spokesperson said that signing only one chapter of the Code is possible – but added that the Commission would have preferred for xAI to sign the other chapters too.
xAI will still have to demonstrate compliance with all of the AI Act, the spokesperson further noted.
X’s chatbot Grok came under fire in the EU earlier this month, after it produced a wave of antisemitic posts and referred to itself as “Mechahitler”.
The outburst led to Poland calling on the Commission to investigate Grok under the bloc’s Digital Services Act.
(nl)
