California's SB 574 mandates strict AI review and limits for lawyers, arbitrators
California's SB 574 requires lawyers to verify AI content and restricts arbitrators' AI decision use.
Why it matters: This law sets concrete legal obligations for AI use by lawyers and arbitrators, affecting compliance and risk management in daily legal practice. It clarifies when and how generative AI can be incorporated responsibly in legal documents and arbitration, helping legal professionals avoid ethical pitfalls and ensure procedural integrity.
- SB 574 was signed by Gov. Gavin Newsom on Sept. 30, 2026, effective Jan. 1, 2027.
- Lawyers must verify and correct AI-generated legal content, including citations, before use and disclose AI assistance in filings.
- Confidential or sensitive information can only be input into AI systems with confined access limited to authorized, confidentiality-bound individuals.
- Arbitrators are prohibited from delegating decisions to AI or relying on AI outputs outside the official record without notifying parties.
California's Governor Gavin Newsom signed Senate Bill 574 (SB 574) into law on September 30, 2026, introducing targeted regulations for generative AI use in legal work and arbitration. Effective January 1, 2027, this legislation imposes specific duties that cement attorneys’ professional accountability when deploying AI tools in their practice.
Under SB 574, lawyers must personally verify the accuracy of any AI-generated legal content, including reviewing case law and statutory citations for correctness before using them in legal documents. This requirement aims to prevent the inclusion of incorrect or fabricated references—often called AI 'hallucinations'—which are unsubstantiated or misleading outputs from AI systems. Attorneys must also disclose the use of AI tools in document preparation in court filings.
The law restricts the input of confidential or nonpublic information into AI systems unless access to such data is tightly controlled and limited to authorized persons who are legally bound to confidentiality. This addresses concerns about data privacy and unauthorized dissemination within AI platforms.
For arbitration, SB 574 prohibits arbitrators from delegating any part of their decision-making to generative AI and bars reliance on AI-generated information not officially included in the arbitration record unless all parties receive prior notice. This ensures transparency and maintains the integrity of arbitration outcomes.
The California Judicial Council is tasked with updating judicial administration standards in light of AI technology, reflecting an adaptive regulatory approach.
As legal experts at Holland & Knight explain, "The statute's foundation is professional accountability: AI may assist, but lawyers remain responsible for all legal work." Independent coverage from the legal news outlet Law360 emphasizes that SB 574 is the first U.S. state law to define clear AI-related duties for legal professionals.
By the numbers:
- 2026-09-30 — SB 574 signed into law by Gov. Newsom
- 2027-01-01 — SB 574 comes into effect
- 0 — AI delegation allowed to arbitrators under the new law
Yes, but: While SB 574 establishes foundational AI rules for California, enforcement details and practical compliance challenges remain to be clarified by courts and regulators.
What's next: The California Judicial Council will revise judicial administration standards to incorporate AI guidelines, expected in late 2027.