California Becomes First State to Legislate Lawyers’ Use of AI

California Becomes First State to Legislate Lawyers’ Use of AI

On Sept. 30, California Gov. Gavin Newsom signed into law a measure regulating lawyers’ use of artificial intelligence in their work. The first-in-the-nation legislation, SB 574, requires lawyers to:

• refrain from entering confidential, personal identifying, and other nonpublic information into a generative artificial intelligence system unless access within the system is restricted to the attorney and authorized persons who are themselves bound by confidentiality obligations;
• take reasonable steps to verify the accuracy of generative artificial intelligence outputs (including the accuracy of all case and statutory citations);
• take reasonable steps to correct erroneous or hallucinated AI materials; and
• disclose the use of generative artificial intelligence in all documents submitted to the court.

SB 574 also amends the California Code of Civil Procedure’s requirements on lawyer signatures on pleadings and other documents filed with the courts. In addition to new rules on the legal effect of a lawyer’s signature on a court filing, Section 128.7(a)(2)(A) of the Code of Civil Procedure now contains the following obligation:

A brief, pleading, motion, or any other paper filed in any court shall not contain any citations that an attorney responsible for submitting the pleading has not personally verified, including any citation provided by generative artificial intelligence.

There are also new rules in SB 574 governing how trial courts should conduct sanctions proceedings that arise from AI misuse. The new legislation appears to address the reputational harm that often accompanies the discovery of AI-generated errors in court filings by providing that AI-related sanctions disputes should be initially worked out among attorneys before they are brought to the court’s attention in a filing on the court’s public docket.

Finally, SB 574 obliges lawyers to “consider whether to disclose” the use of generative artificial intelligence in content shared with the public.

It would be reasonable to expect that other states will be looking closely at California’s approach to regulating artificial intelligence. The state is often a first-mover in the area of technology-related policymaking. Earlier this year, the State Bar of California proposed amendments to six Rules of Professional Conduct that would make AI obligations enforceable through discipline.

Developing evidence rules for AI-generated materials may prove far more challenging for courts and policymakers.

Evidence Rules Forthcoming?

Regulation of lawyers’ use of artificial intelligence is one matter where consensus views on what constitutes unacceptable behavior have been easy to reach. Lawyers should be competent in the use of technology, of course; and pleadings filed with the court should be written with care, containing accurate statements of law and no obvious falsehoods.

Developing evidence rules for AI-generated materials may prove far more challenging for courts and policymakers. Trial judges across the country, working with pre-AI evidentiary rules, are increasingly encountering evidence that has been created or manipulated by artificial intelligence. The National Center for State Courts, an educational clearinghouse for state court policymakers, recently urged action on evidence rules governing AI-generated materials, arguing that deepfake evidence threatens to undermine public trust in the legal system. But so far, only Louisiana has written evidence rules specifically addressing authenticity issues when AI-generated evidence is presented in court. Article 371 of the Louisiana Code of Civil Procedure requires lawyers to use “reasonable diligence” to verify the authenticity of evidence before offering it into evidence. Article 1551 requires parties to disclose at the pretrial conference whether their evidence has been generated by artificial intelligence and whether they suspect an opposing exhibit was likewise fashioned by AI.

At the federal level, the U.S. Judicial Conference’s Committee on Rules of Practice and Procedure has proposed Federal Rule of Evidence 707, which would require a reliability determination before AI-generated evidence could be admitted.

California may join the fight soon. The California Judicial Council’s Artificial Intelligence Task Force has been asked to develop guidance for the use of AI-generated evidence in court. Legislation directing the council to prepare a report on AI-generated evidence (SB 11) by Jan. 1, 2027, passed the California legislature but drew the governor’s veto in October 2025. Nevertheless, the judicial council appears to be moving ahead on AI-generated evidence policy without the legislature’s explicit insistence.

Finally, outside of the courtroom (though just slightly), the rapid emergence of technologies that can create false but convincing evidence is a topic of strong interest among fraud regulators. A 2025 paper presented to the SEC Investor Advisory Committee, AI, Deepfakes, and the Future of Financial Deception, documented how the technology to create false evidence is advancing well ahead of human beings’ ability to detect inauthenticity. Computers, including those with bad intent, learn more quickly than humans.