Elon Musk's xAI has failed to block California's new law requiring AI companies to disclose their training data sources. US District Judge Jesus Bernal rejected xAI's request for a preliminary injunction, ruling that the company failed to prove the law would reveal trade secrets or cause irreparable harm.

California's Assembly Bill 2013, which took effect in January, mandates that AI developers explain which datasets were used to train their models, when data was collected, and whether it includes copyrighted or personal information. xAI argued these disclosures would be "economically devastating" and reduce its trade secrets to zero, but the judge found these claims too vague and speculative.

Bernal also dismissed xAI's constitutional arguments, stating the company hadn't demonstrated its datasets or cleaning methods were sufficiently unique to warrant trade secret protection. The judge emphasized that the public has a legitimate interest in understanding how AI models are trained, noting consumers might want to know if medical data or scientific information was used. While xAI's lawsuit continues, the company must now comply with the disclosure requirements, potentially sharing information with competitors like OpenAI that Musk has been battling in court.