California has moved ahead of Washington on Artificial Intelligence policy, with Governor Gavin Newsom signing a package of 18 laws that together create the most comprehensive state framework to date. The centerpiece is the Transparency in Frontier Artificial Intelligence Act, formerly Senate Bill 53, which targets developers of the most powerful systems and positions California as the first state to directly regulate Artificial Intelligence safety. With federal preemption efforts stalled, the state’s approach is likely to influence national practice, similar to how the California Consumer Privacy Act set a baseline for privacy standards. Many of the new measures took effect in January 2025, with additional rules arriving this fall.
The flagship law requires frontier model developers to publish safety plans that align with national and international standards, report critical incidents to the California Office of Emergency Services, protect whistleblowers who raise health and safety concerns, and support public research via CalCompute, a new state-backed computing consortium for developing safe and ethical Artificial Intelligence. After intensive industry lobbying, the final measure softened some provisions: incident reporting was narrowed to events causing physical harm, and penalties were reduced compared with earlier drafts, raising concerns that large companies may see violations as a manageable cost of doing business.
Beyond safety, the package reshapes use of Artificial Intelligence across sectors. On election integrity, SB 926 criminalizes nonconsensual deepfake pornography, while laws such as AB 2655 and AB 2355 require platforms to label or block materially deceptive election content, including Artificial Intelligence-generated audio and video, ahead of the 2026 cycle. On privacy, AB 1008 designates Artificial Intelligence-generated profiles, predictions, and inferences as personal information under the California Consumer Privacy Act, extending access, deletion, and control rights. SB 1223 adds heightened protection for neural data by classifying it as sensitive personal information.
Workplace rules are also tightening. New regulations from California’s Civil Rights Department, effective October 1, 2025, bar the use of automated decision systems that produce discriminatory outcomes under the Fair Employment and Housing Act. Employers are expected to conduct bias audits of Artificial Intelligence tools used for hiring, promotion, and evaluation, and liability is shifting from intent to impact. In healthcare, SB 1120 prohibits Artificial Intelligence from independently determining medical necessity in insurance utilization reviews, requiring a physician to make the final decision. For businesses, immediate steps include documenting safety practices, bias testing, and whistleblower protections; vetting vendors on audits, liability, and privacy compliance; and, for healthcare entities, ensuring physician oversight and clear human review procedures. The analysis is current as of September 30, 2025.
