California moves to police lawyers’ use of AI agents, signaling tougher rules for professionals

California has enacted a new law, SB 574, that specifically targets how lawyers can use AI systems—especially autonomous “agentic” tools that can act on instructions by accessing external services or sending documents without constant human input. The statute sets out clearer rules on accountability, confidentiality and oversight when generative AI is woven into legal workflows.

Under the law, attorneys remain fully responsible for any filings or contracts drafted with AI assistance. They must disclose AI use to clients where relevant, maintain records of how tools were used, and cannot delegate high‑risk professional judgments entirely to automated systems. The legislation also nudges firms to build review processes that check for bias and factual errors in AI‑generated work.

Because California hosts many of the world’s largest tech companies, its approach to AI often becomes an informal standard for other US states. Observers expect similar guardrails for doctors, financial advisers and other licensed professionals, as regulators try to balance innovation with consumer protection. At the same time, some startups warn that overly rigid rules could slow experimentation with AI agents that handle routine, low‑stakes tasks in legal practice.