California’s approval of SB 574 is emerging as one of the most consequential legal-industry developments of the year, even without a same-day court ruling attached to it. The bill would place clear limits on how lawyers use generative AI in legal work, including restricting delegation of legal practice to AI tools, requiring verification of AI-generated material, and imposing disclosure-related constraints in court filings.
At a high level, the measure reflects a simple but increasingly urgent principle: lawyers can use AI, but they cannot outsource professional judgment to it. That distinction has been at the center of growing judicial and bar-level concern after a series of well-publicized incidents involving fabricated citations, inaccurate factual assertions, and AI-assisted drafting that slipped into court submissions without adequate review.
For litigators, the practical implications are immediate. If enacted and implemented as expected, SB 574 would reinforce that attorneys remain fully responsible for the accuracy of pleadings, motions, briefs, and supporting authorities, regardless of whether an AI platform helped draft them. Verification obligations could also affect law firm workflows, requiring more formal review processes before filings are submitted. In high-volume litigation practices, that may mean new internal controls around cite-checking, factual validation, and documentation of attorney supervision.
In-house counsel and compliance teams should also pay close attention. Many legal departments are actively piloting generative AI for contract review, internal investigations, regulatory responses, and outside counsel management. A California framework that expressly addresses attorney use of AI could become a template for corporate governance policies, vendor diligence standards, and training programs. Even organizations operating outside California may find themselves adapting to its approach if their lawyers practice in the state or appear in California courts.
The broader significance is that SB 574 may help shape the next phase of professional-responsibility rules nationwide. Courts and regulators have so far taken a patchwork approach to AI, relying on standing orders, ethics opinions, and existing duties of competence, candor, and supervision. California’s move suggests a more direct legislative model—one that spells out what lawyers may not delegate and what must be independently confirmed.
That makes this more than a state policy story. It is a signal that the legal profession is entering a more regulated phase of AI adoption, where efficiency gains will be judged against core duties to clients, courts, and the integrity of the legal system. For legal professionals, the message is becoming harder to ignore: experimentation with AI may still be encouraged, but unverified reliance on it increasingly will not be tolerated.