OpenAI told the California Assembly’s Privacy and Consumer Protection Committee on Oct. 27 that it supports Senate Bill 53 but wants the measure clarified to require stronger safeguards for highly capable AI systems. The company’s testimony is documented in the committee hearing audio.
Among the changes OpenAI proposed were effective “shutdown controls” that could pause training, isolate AI agents, revoke tool or network access, or suspend services when necessary, according to the testimony. The company also called for prompt written notice to organizations whose protected systems or confidential information are materially accessed, altered or destroyed during a serious cyber incident.
OpenAI further supported independent technical assessments covering areas such as alignment, containment and monitoring. The company said California could build on existing state measures, including SB 813 and AB 1405, as it develops standards for independent assessments.
The hearing also included testimony from other technology companies and AI-safety researchers about model evaluations, cybersecurity incidents and safeguards. But the supplied hearing record does not show a formal motion, vote or other committee action on SB 53. The bill’s current text, legislative status and whether lawmakers will pursue amendments were not established in the materials reviewed.
