California has enacted what Reuters describes as the first US state law specifically regulating lawyers’ and arbitrators’ use of generative AI.
SB 574 prohibits lawyers from delegating the practice of law to AI. Attorneys using it must take reasonable steps to verify outputs, correct erroneous or hallucinated material and disclose AI use in documents submitted to a court. An attorney responsible for a filing must personally verify its citations. The law also restricts entering confidential or non-public information into systems where access is not appropriately controlled.
Arbitrators may use AI as an aid but cannot delegate any part of their decision-making or rely on generated information outside the record without appropriate disclosure. They remain responsible for the resulting award.
Many of these requirements resemble existing professional duties. The consequential change is that California has converted them into express statutory boundaries. AI assistance is permitted, but legal judgment, verification and responsibility must remain human.
For Australian practitioners, the direction is familiar: courts already require verification and, in some circumstances, disclosure of AI-assisted material. California’s approach suggests that future regulation may go further by expressly separating machine assistance from professional adoption. Legal technology will therefore need to record not only which sources were checked, but who reviewed the reasoning, corrected it and accepted responsibility for the final work.
Sources: California announcement; SB 574; Reuters.
Leave a Reply