Disclosure decisions arise if California lawyers use generative AI

California Governor Gavin Newsom signed into law first-of-its-kind state legislation setting rules for how lawyers are permitted to use generative AI, Reuters Legal reports. Legaltech News reports that the law, SB 574, gives lawyers stronger AI review responsibilities and bars arbitrators from relinquishing their decision-making powers to generative artificial intelligence. The law was signed in California alongside other AI-related measures, according to Reuters Legal.

The significance is not that California has declared generative AI unusable in legal work. It is that lawyer use remains tied to human responsibility. The law places review responsibilities on lawyers and creates disclosure decisions for firms determining when their use of the technology requires disclosure, according to Legaltech News and Bloomberg Law. That moves the practical question away from whether a system can produce text and toward whether the legal work remains subject to accountable review. For an AI-native legal model, that distinction is central: speed or automation does not finish the work if the lawyer remains responsible for what reaches a client, opposing party, court or arbitrator.

The signing establishes a state rule governing lawyers’ permitted use of generative AI. It does not establish that review under the law will be effective, that firms will make disclosure decisions consistently, or that AI-generated legal work will become reliable merely because a lawyer reviews it. Nor does the arbitrator provision show that every legal task requires the same form of human involvement; it shows only that arbitrators may not relinquish their decision-making powers to generative artificial intelligence under California law. A stronger reading would require the law to be applied in an actual dispute or disclosure decision, with a consequence attached to inadequate review or an omitted disclosure. Nothing in the reported signing establishes that application. The event therefore supports a narrower but important proposition: AI use does not remove the consequences of bad legal work, and California has put review and disclosure at the center of the rule governing that use.

The next fact to watch is a published California decision or disclosure determination applying SB 574 to a lawyer’s use of generative AI. The useful signal will be whether the decision identifies what review was required, when disclosure was required and what consequence followed when either was missing. A further test is whether California’s bar, courts or arbitrators apply the law to the quality of completed legal work rather than treating disclosure alone as the endpoint. We will count the law as meaningful evidence for accountable AI legal work only when that application is visible; until then, the signing is a rulemaking event, not proof that the underlying work has become safer.