Overview
- SB 53, signed by Gov. Gavin Newsom, compels leading AI developers to publish safety and security protocols and to report critical incidents to California’s Office of Emergency Services within 15 days, with faster reporting if there is imminent harm.
- Covered firms include those with more than $500 million in annual revenue and models trained at very high compute levels, concentrating requirements on companies such as OpenAI, Google DeepMind, Meta, and Anthropic while exempting most startups.
- Reportable events include model‑enabled weapons assistance, major cyberattacks, loss of control, and dangerous deceptive behavior observed in testing, expanding visibility into high‑risk AI failures.
- Enforcement includes civil penalties of up to $1 million by the state attorney general, and the law strengthens whistleblower protections by barring retaliation and requiring companies to inform employees how to report concerns.
- The law takes effect Jan. 1, 2026, establishes the CalCompute public cloud with a framework due by Jan. 1, 2027, and arrives as industry reaction splits and federal lawmakers float a parallel DOE evaluation program.