California moves from guidance to legislation
Senate Bill 574, signed into California law on September 30 and effective January 1, 2027, creates specific obligations for lawyers using generative AI.
Lawyers remain responsible for AI-generated work
Among other things, lawyers cannot delegate the practice of law to AI and must take reasonable steps to verify AI-generated output and correct erroneous or hallucinated content.
Confidentiality and verification take centre stage
The legislation also restricts lawyers from entering confidential, identifying and other non-public information into AI systems without appropriate access and confidentiality protections and requires lawyers to verify citations in court filings, including citations generated by AI. Failure to comply can have real consequences, including court-imposed penalties and sanctions.
What Canadian law firms should take from this
As AI becomes part of everyday legal practice, law firms and in-house legal departments need to think beyond whether lawyers are permitted to use it and focus on the controls around how it is used: protecting confidential information, verifying output, ensuring appropriate lawyer review and maintaining responsibility for the finished product.
AI can assist, but accountability remains human
California may have chosen legislation where Canadian regulators have so far largely taken a guidance-based approach, but the underlying principle is increasingly familiar: AI can assist the lawyer, but it cannot assume the lawyer's professional judgment or responsibility.
/Passle/6a705cc2ad7be9cbfe860544/SearchServiceImages/2026-10-06-14-19-55-025-6ac5038be14e834091bd80f2.jpg)
/Passle/6a705cc2ad7be9cbfe860544/SearchServiceImages/2026-09-29-23-30-57-442-6abc4a31d850e7795f857d31.jpg)
/Passle/6a705cc2ad7be9cbfe860544/SearchServiceImages/2026-10-08-02-40-46-651-6ac702ae8f60cb75fde0e1ac.jpg)