What xAI’s Challenge to California’s Generative AI Disclosure Law Means for Trade Secrets

Lexology PRO reported on xAI’s lawsuit seeking to block enforcement of California’s new AI transparency law, AB 2013, which requires generative AI developers to publish a “highlevel summary” of their training datasets. xAI argues the disclosures would force the company to reveal protected trade secrets, calling the law unconstitutional.  

Finnegan partner Karthik Kumar, who is not involved in the litigation, noted that the law’s requirements may indeed sweep in sensitive proprietary information. He explained that “some of the provisions in the law could potentially loop in companies’ trade secrets.” Karthik added that disclosures about dataset cleaning, processing, and other backend steps could expose competitive strategy, emphasizing that “having to disclose some of those aspects could involve [showing] what gives you a competitive advantage.” 

Karthik also observed that the statute’s vagueness creates compliance challenges, including uncertainty around how much detail is required in the mandated summaries. Looking ahead, he noted that if AB 2013 survives judicial scrutiny, AI developers may shift toward stronger IP protection strategies, saying “it would certainly force in-house counsel to consider the possibility of patenting more aggressively.” 

The lawsuit underscores growing tension between transparency mandates and the protection of proprietary AI development practices, as companies, trade groups, and policymakers continue to debate how to balance innovation, competition, and consumer protection. 

Read What xAI’s Challenge to California’s Generative AI Disclosure Law Means for Trade Secrets