California Legislature · 2025

California SB 53 — Transparency in Frontier AI Act

CA SB 53

California's evolution from the vetoed SB 1047 to SB 53 illustrates the real-time negotiation between ambition and political feasibility in AI governance. SB 53 targets large frontier developers with over $500M annual revenue and requires transparency reports on safety testing, with critical safety incident reporting within 15 days standard or 24 hours if imminent harm.

Key Provisions

Regulatory Philosophy

Iterative, transparency-driven, with voluntary alignment toward federal standards. California's shift from SB 1047's mandate-heavy approach to SB 53's transparency-first model reflects the political difficulty of imposing pre-deployment requirements on frontier developers headquartered in the state. Disclosure over prohibition is the guiding principle.

Strengths

Derived from the proposal’s own policy documents

  • +Transparency-first approach is politically achievable where SB 1047's mandate-heavy predecessor was not — iterative governance that can actually be implemented
  • +The $500M revenue threshold targets only the largest frontier developers, avoiding burdens on startups and academic researchers
  • +Critical safety incident reporting with a 24-hour window for imminent harm creates real accountability for the most dangerous failures
  • +Alignment toward CAISI standards signals willingness to harmonize with federal frameworks, reducing fragmentation risk
  • +Demonstrates that state-level AI governance can evolve in response to industry and political feedback without abandoning the regulatory project

Weaknesses

From the perspective of political opposition

  • −SB 53 is the tombstone of SB 1047 — California had the chance to set a real safety standard and instead caved to industry lobbying, producing a disclosure regime with no enforcement teeth
  • −Transparency reports are a compliance theater — companies will disclose what makes them look good and obscure what does not, and there is no mechanism to verify accuracy
  • −No pre-deployment testing, no kill switch, no liability — SB 53 asks companies to tell us about dangerous AI after they have already released it to the public
  • −The $500M threshold conveniently exempts every AI startup in the state, creating a loophole large enough to drive a frontier model through
  • −By retreating from SB 1047, California signaled to every other state that industry pressure works — making meaningful state-level regulation harder everywhere

Position on Analytical Frameworks

Enforcement Mechanism vs. Regulatory Scope

Prevention vs. Liability & Regulatory Authority

Innovation Priority vs. Worker Protection

Pre-deployment Obligations vs. Federal Preemption

← Back to all proposals