Should California’s AI law demand disclosure, accountability survives automation

California Governor Gavin Newsom signed legislation on October 1, 2026, setting rules for how lawyers may use generative AI in their work, according to Reuters Legal. Bloomberg Law, Business & Practice reports that the measure is a first-of-its-kind bill on lawyer and arbitrator AI use. Legaltech News (Law.com) reports that SB 574 imposes stronger AI review responsibilities on lawyers and bars arbitrators from relinquishing their decision-making powers to generative artificial intelligence.

The significance is not that California has made generative AI unusable. It is that California law places the use of the technology inside ordinary professional accountability. A lawyer’s use of AI now sits alongside review responsibilities, while a separate rule keeps arbitrators from handing decision-making power to a machine. Bloomberg Law, Business & Practice also reports that the law gives firms difficult disclosure decisions about when AI use must be revealed. For an AI-native legal business, that is a direct test of whether its process can produce work that survives human review and any disclosure duty attached to it. Automation changes how work is produced; it does not remove the consequences when the work is defective.

The signing establishes that California has enacted rules governing lawyers’ use of generative AI. It does not establish that the rules will identify every defective citation, unsupported assertion, confidentiality failure or other problem in AI-assisted legal work. It also does not establish that a firm can satisfy its obligations merely by disclosing that it used the technology. The stronger reading would require more than the existence of duties: it would require a demonstrated connection between the duties, the firm’s review process and the quality of the work delivered to a client or filed in a proceeding. Nothing in the reported signing supplies that connection. The order of responsibility remains clear, however. California law regulates the lawyer and the arbitrator, not an abstract promise made by a model. A lawyer remains the accountable actor for legal work, and an arbitrator remains responsible for the decision the law does not permit artificial intelligence to make in the arbitrator’s place. That is a narrow result, but it is the result the event supports.

The next confirming fact would be a California court or disciplinary order applying the review or disclosure duties to work produced with generative AI. Such an order would show whether the law reaches beyond formal compliance and attaches consequences to deficient work. A filing that treats disclosure as sufficient even where the work is wrong would cut against the stronger accountability reading. We will watch for an order, filing or reported disclosure decision under SB 574 that identifies what review must accomplish and who bears the consequence when it does not.