Judge denies xAI's bid to block California's AI training-data transparency law

What happened

U.S. District Judge Jesus G. Bernal denied xAI's motion for a preliminary injunction on 4 March 2026, refusing to block enforcement of California's AI training-data transparency law, Assembly Bill 2013, while xAI's constitutional challenge proceeds (X.AI LLC v. Bonta, No. 2:25-cv-12295, C.D. Cal.).

What AB 2013 requires

AB 2013, signed in September 2024 and effective 1 January 2026, requires developers of generative AI systems publicly available to Californians to post documentation on the data used to train their systems, including a high-level summary covering at least 12 topics: the sources or owners of datasets, the number and types of data points, whether datasets include copyright-protected material, whether datasets were purchased or licensed, whether personal information is present, cleaning or processing applied, collection time periods, and whether synthetic data was used. Its scope reaches models released since 2022.

The ruling

xAI sued Attorney General Rob Bonta on 29 December 2025, arguing the law violates the Takings Clause, the First Amendment, and is unconstitutionally vague. Judge Bernal found xAI had standing to sue but had not shown a likelihood of success on any of the three claims: its trade-secret pleadings were too general to show its datasets qualify for protection; AB 2013 likely regulates commercial speech, subject to intermediate scrutiny rather than strict scrutiny; and the vagueness challenge was premature on an underdeveloped record. The court noted xAI showed only a "distinct possibility" of prevailing on its First Amendment claim, which is not the likelihood of success a preliminary injunction requires.

What comes next

The denial is interim relief only and does not decide the case. xAI appealed to the Ninth Circuit (No. 26-1591), with oral argument scheduled for 18 November 2026.