Appeals Judges Skeptical of xAI Bid to Block California AI Data Law

A federal appeals court gave Elon Musk's xAI what Bloomberg Law described as "a cold reception" on Wednesday, Nov. 18, as the company tried to block a California law forcing AI developers to disclose what data trains their models. The US Court of Appeals for the Ninth Circuit has not issued a decision, and the panel pressed both xAI and the state, according to MLex.
The fight centers on Assembly Bill 2013, which took effect on Jan. 1, 2026. It requires developers of publicly available generative AI systems to post a high-level summary of their training data — at least 12 topics, including where datasets came from, whether they contain copyrighted material or personal information, and whether the data was bought or licensed.
What did the judges actually say?
Judge Johnnie B. Rawlinson challenged xAI's core argument that the law unconstitutionally regulates commercial speech. "Isn't it reasonable for the government to want its citizens to be able to know upfront what information is being input into the AI instrument," she asked, according to Bloomberg Law. That question cuts at the heart of xAI's First Amendment theory. If the court sees the disclosures as reasonable factual transparency rather than compelled ideological speech, the claim weakens considerably.
The panel didn't let the state off easy either. MLex reported that judges pressed California's side on the law's scope and burdens, suggesting the court is still weighing where the disclosure line should sit.
How did the case get here?
The timeline matters for understanding the stakes:
- Dec. 29, 2025 — xAI sued California Attorney General Rob Bonta, arguing the law takes its trade secrets without compensation, compels speech in violation of the First Amendment, and is unconstitutionally vague.
- March 4 — US District Judge Jesus Bernal denied a preliminary injunction, finding xAI had not shown it was likely to succeed on the merits.
- Nov. 18 — The Ninth Circuit heard xAI's appeal of that denial.
Bernal's March ruling was the first real test of AB 2013 in court, and xAI lost it. The appeals argument was the company's second swing, and the reported reception suggests the odds haven't improved.
Why does one California law matter so much?
California hosts most of the major US AI developers, so a disclosure mandate there effectively sets a national baseline. If xAI must publish a twelve-topic summary of its training data — sources, licensing status, copyrighted content, personal information — competitors building publicly available generative systems face the same rules.
Three legal theories are in play. The trade-secrets claim argues the state is taking proprietary information without compensation. The First Amendment claim frames the disclosures as compelled speech. The vagueness claim says the law's requirements are too ill-defined to follow with confidence. The district court found none of them strong enough to warrant an injunction, and Rawlinson's questioning on Nov. 18 signaled at least one appellate judge sees the disclosure requirement as reasonable regulation.
What comes next?
No ruling date is set, and AB 2013 remains in force while the appeal is pending — meaning developers are already expected to comply with the disclosure requirements xAI is contesting. A Ninth Circuit decision upholding Bernal's denial would leave xAI with a Supreme Court petition as its main remaining option, while a reversal would reopen the question of whether training-data transparency can be compelled at all.
Source: mlq.ai
Rebecca Stone
Show full bio
Correspondent covering technology and product news in Semiconductors, chip manufacturing, hardware, AI compute at Chip Dispatch.
