OpenAI proposes six amendments to California's SB 53, covering RSI, shutdown controls, audits
ShakeelHashim · x · 2026-10-22
OpenAI has written to Governor Gavin Newsom recommending a package of amendments to strengthen SB 53, calling the frontier AI safety law a strong start that must evolve as capabilities advance, particularly around automated AI R&D.
Key proposals:
- Explicitly address automated AI research and recursive self-improvement (RSI) in safety frameworks, assessing whether such capabilities raise catastrophic risk or make it harder to monitor
- Stronger internal cybersecurity to protect unreleased model weights, including sandboxing models from internet access during training and evaluation
- Required monitoring of unsafe model behavior, including attempts to circumvent safeguards, tied to clear escalation and incident-response procedures
- Written notice to third parties affected by cyber incidents involving frontier models
- Effective shutdown controls, with California establishing enforceable halt conditions via rulemaking absent national standards
- Mandatory annual independent compliance audits
OpenAI also notes its Frontier Governance Framework already reports quarterly to California's Office of Emergency Services on catastrophic risks from internal model use.
More from Safety
- Researcher warns GitLab flaw is 'really bad,' urges users to patch now — basedjensen · 2026-10-22
- RCE free-for-all feels like 2000 again, vet says, pointing to PHP2Shell race bugs — moyix · 2026-10-22
- Microsoft: Storm-0940 used agentic AI framework ARTEX to attack hundreds of orgs — basedjensen · 2026-10-22
- Amid rising cyberattacks, a 6-step personal security checklist worth doing today — peterwildeford · 2026-10-22
- GPT-5.6 Sol and Gemini 4 Argon both found a critical PHP flaw affecting millions of sites — The Decoder · 2026-10-22
- 136-Country, 118M-Article Dataset Audits State Media Share in LLM Training Corpora — deliprao · 2026-10-22