Governor Gavin Newsom signed SB 1000 (Becker, D-Menlo Park) on September 30 as an urgency statute, deleting the 1,000,000-monthly-user floor that until then had kept the California AI Transparency Act aimed squarely at the frontier labs. The bill, chaptered as Chapter 861, took effect the instant it was signed. Every publicly accessible generative AI system touching California users is now in scope.

That’s the structural move worth sitting with. CATA was enacted in 2024 via SB 942, amended in 2025 by AB 853, and only reached effect in August. The million-user threshold was the political compromise that made it palatable. SB 1000 pulls it out, and the media and entertainment carve-out now covers only non-user-generated video games, per Morrison Foerster’s reading of the chaptered text.

What founders of 5-to-30-person shops now owe California: a disclosure-verification tool (replacing the old “AI detection tool,” and satisfiable through a qualifying third-party option, excluding minor edits like cropping or resizing), and a latent provenance disclosure that must indicate whether the system created or altered the content. The manifest-disclosure option is gone. Beginning January 1, 2029, systems must also disclose whether they’re designed primarily as assistive technology, the category temporarily exempt until that same date.

Penalties bite daily. $5,000 per violation per day, with each day a discrete violation and prevailing plaintiffs collecting attorney’s fees. Falsely claiming the assistive-technology exemption runs $50,000 per day. A new licensee framework gives providers 72 hours to terminate or notify, and licensees 96 hours to comply or cease, with unresolved cases escalated to the California Attorney General.

SB 1000 arrived alongside SB 947, AB 2713, and SB 1159 in the roughly dozen measures Newsom signed that day, closing the window that had been open hours earlier. The size exemption, in retrospect, was always borrowed time.

Sources

Sources