Particle.news

Download on the App Store

California Enacts First-in-Nation Frontier AI Law Requiring Big Labs to Disclose Safety Protocols

Supporters frame it as a template for national rules by focusing on only the highest‑capacity developers.

Overview

  • Governor Gavin Newsom signed SB 53 on September 29, creating transparency and reporting rules for “large frontier developers” that both train models above 10^26 operations and generate more than $500 million in annual revenue.
  • Covered companies must publish a frontier AI framework detailing governance, catastrophic‑risk thresholds and mitigations, cybersecurity practices, and evaluation plans, and issue public transparency reports before deploying new or substantially modified models.
  • Developers must report critical safety incidents to California’s Office of Emergency Services within 15 days, with 24‑hour notification to appropriate authorities if imminent harm is identified, and certain reports are shielded from public disclosure.
  • The law includes whistleblower protections requiring anonymous internal reporting channels and barring retaliation, and it authorizes civil penalties up to $1 million per violation enforced by the California Attorney General starting January 1, 2026.
  • SB 53 establishes CalCompute, a University of California‑run public cloud initiative for research and startups, while industry reaction is mixed and officials cast the measure as a potential national standard as federal policy remains unsettled.