Anthropic is endorsing SB 53

Source: Anthropic Date: 2025-09-08 URL: https://www.anthropic.com/news/anthropic-is-endorsing-sb-53

Summary

Anthropic endorsed California’s SB 53 — legislation requiring frontier AI developers to disclose safety frameworks, publish transparency reports on catastrophic risk assessments, report critical safety incidents within 15 days, and implement whistleblower protections. Anthropic’s rationale: SB 53 formalizes what it already does (RSP, system cards) and creates a level playing field. Key framing: “AI advancements won’t wait for consensus in Washington” — California stepping in where federal regulation has stalled.

Implications

• Policy / regulatory strategy. Endorsing SB 53 is the opposite of the industry’s opposition to California’s 2024 SB 1047 (which Anthropic also opposed). The difference: SB 53 uses disclosure requirements rather than prescriptive technical mandates. Anthropic’s consistent position is: disclose-and-verify frameworks good, technical-requirements-in-legislation bad.
• Level playing field argument. The competitive argument (mandatory disclosure prevents competitors gaining advantage by reducing safety transparency) is Anthropic’s core regulatory strategy position — it benefits from safety requirements that it already meets.
• 15-day incident reporting. The 15-day critical safety incident reporting requirement is the operationally significant provision — it creates a reporting obligation that forces disclosure of incidents that companies might otherwise handle quietly.
• SB 53 vs. SB 1047 comparison. SB 1047 (which the governor vetoed in 2024) created liability for AI harms and required third-party audits — prescriptive and costly. SB 53 requires disclosure only. Anthropic’s SB 1047 opposition + SB 53 endorsement defines its legislative preference function precisely.
• Watch: whether SB 53 was enacted and how it was implemented; whether the 15-day incident reporting produced any notable public disclosures; whether other states adopted similar frameworks.