Ninth Circuit hears xAI challenge to California AI training-data disclosure law
|
The facts
- The Ninth Circuit heard xAI's request to block California's AI training-data disclosure law.
- California's law requires generative AI companies to publish summaries of training datasets.
- xAI argues the disclosure requirement violates the First Amendment.
- xAI says the requirement could expose proprietary training information.
- The appellate judges questioned the breadth of disclosures California could require.
- California's AI training-data disclosure law took effect January 1.
- A federal judge previously denied xAI's request for an injunction against the law.
Context
What does California's law require?
The law requires generative AI companies to publicly post summaries of the datasets used to train their AI systems. Reuters Startup Fortune Cryptopolitan
Why is xAI challenging the law?
xAI argues that mandated disclosures compel speech in violation of the First Amendment and could reveal proprietary details about its model-training process. TokenPost Startup Fortune Cryptopolitan
What remains unresolved?
The Ninth Circuit has not ruled on xAI's appeal, and the hearing included questions about both xAI's constitutional argument and the scope of information California may demand. Reuters Startup Fortune Cryptopolitan
Common ground, and where they split
- Common ground
- Both frames accept that California's law compels generative AI firms to publish training-data summaries and that xAI's First Amendment challenge turns on how far that compulsion reaches.
- Where Left and Right split
- The left and the right split on whether disclosure serves the public or compels xAI's speech without cause.
- Why they differ
- The split is a values clash over the default burden: the left treats public disclosure as the baseline companies must justify withholding from, while the right treats proprietary secrecy as the baseline the state must justify overriding.
- Watch for
- Watch for the Ninth Circuit panel's ruling, which carries no set date but will come from Judges Rawlinson, Smith and Forrest after the November 18 hearing.Startup Fortune
How left and right read it
The public should not have to take an AI company's word for what it built its systems on. California's law simply asks generative AI firms to publish summaries of their training data, and a federal judge already declined to block it before xAI ever reached this court. Keep the law running.
The default should be that a company keeps what it built to itself, and a state that wants to override that owes a precise justification first. xAI's warning that the requirement could expose proprietary training information goes to that burden, because a mandate to publish summaries of training datasets is a command to speak. The judges pressed California on how broad those disclosures could run. That question comes first.
The disclosure fight really turns on who must justify themselves first — California's mandate, or xAI's silence.
The receipts — 6 sources
Wire services (1)
Other coverage (5)
A news briefing built to show both sides.
The day’s biggest stories in one short brief, built to show both sides and the competing values behind the headlines. Free in your inbox.