It’s one of the first states in the US to do so, Holland & Knight notes
California Governor Gavin Newsom recently approved Senate Bill No. 574 (SB 574), which prohibits the state’s attorneys from delegating the practice of law to generative artificial intelligence (genAI), but not from using such tools to assist in their legal work.
In an alert published shortly after the bill’s signing, Philippa J. Balestrieri, David Jargiello, Kristin A. Ornstein, and Tom Stephenson of Holland & Knight highlighted that California will be one of the first states in the US to explicitly regulate lawyers’ genAI use as of Jan. 1, 2027.
“The statute's premise is professional accountability: AI may assist counsel, but responsibility for the legal work remains with the attorney,” stated the global law firm’s update.
Requirements of California bill on lawyers’ generative AI use
Regarding the prohibition against the delegation of the practice of law, Holland & Knight’s alert stated: “A reasonable interpretation is that AI can assist with research, drafting, document evaluation, discovery review, correspondence and similar work, but attorneys remain responsible for the legal judgment and analysis underlying the work product.”
SB 574 applies to California’s attorneys, arbitrators, judicial officers, and alternative resolution providers. Under the bill, attorneys in the state should:
- Refrain from entering confidential, personal identifying, and other nonpublic information into a genAI system
- Take reasonable steps to verify the accuracy of genAI output
- Make reasonable efforts to correct any erroneous or hallucinated output in the material they use
- Keep from filing in court a brief, pleading, motion, or another paper with AI-generated and other citations that are not personally verified
According to the Holland & Knight alert, the bill’s confidentiality requirement might mean that California’s lawyers cannot utilize some publicly available consumer AI tools in their legal practice.
On the requirement to verify genAI output, the firm’s update said: “Reasonable verification may require confirming that a cited authority exists, remains good law and supports the proposition for which it is cited.”
Regarding the verification of court filings, the Holland & Knight alert called attention to the breadth of this requirement, noting that “It creates a personal verification requirement for every citation in a California court filing, whether from AI, an associate, a paralegal, research service, prior brief or another source.”
- delegating any part of their decision-making process to genAI
- relying on AI-generated information outside the record without prior disclosure to the parties
Under SB 574, the Judicial Council established by the California Constitution should publicly revisit the relevant judicial administration standard to ensure that it incorporates any changes needed to reflect the future development of genAI.
More international law news on artificial intelligence
Here are some other recent international law news stories revolving around AI.
Robert Buckland, a former UK lord chancellor, encouraged judges to utilize the costs regime to limit the increase of “AI slop” litigation brought by litigants in person, the Law Society Gazette reported.
Paying subscribers to AI chatbots ChatGPT, Claude, Grok, and Gemini sued AI giants Anthropic, OpenAI, SpaceXAI, and Google for violating antitrust law by jointly agreeing to decelerate AI development, The Associated Press reported.
Enjoy this story? Read the latest international law news on the main page!
