XAI Broke Its Safety Policy With Coding Model

California Gov. Gavin Newsom signed SB 53 on Monday, a law that will require the largest AI companies to abide by their own safety and cybersecurity policies. One company that should take heed of the new law is Elon Musk's xAI.

The ink had barely dried on xAI’s safety policy, which it published in August, when the company released its model Grok Code Fast 1 and violated the policy’s terms, according to company documents. AI Lab Watch, an industry watchdog that tracks the safety practices of leading AI companies, recently pointed out the violation, which has not been previously reported.

“It's concerning that xAI is violating the safety policy, and trying to do so quietly,” said Zach Stein-Perlman, who runs AI Lab Watch.