Overview

• Gov. Gavin Newsom signed SB 53 on Monday, creating the nation’s first state law focused on safety oversight of cutting‑edge AI models, effective January 1, 2026.
• The law applies to the largest AI developers that meet thresholds such as more than $500 million in annual revenue and training runs around 10^26 FLOPs.
• Companies must publish safety and security frameworks and report critical incidents to California’s Office of Emergency Services within 15 days, with anonymized public summaries starting in 2027.
• The measure strengthens whistleblower protections and empowers the California Attorney General to seek civil penalties of up to $1 million per violation.
• Requirements include disclosing dangerous deceptive behavior observed in testing, and the law follows a 2024 veto that spurred an expert working group; industry reaction is mixed, with Anthropic supportive and others urging federal rules.