California Governor Gavin Newsom announced a new proposal on Wednesday aimed at enhancing the oversight of artificial intelligence in the state. Described as a ‘nation-leading’ framework, the plan would enable authorities to intervene and shut down rogue AI systems in emergencies.
The proposal follows a lawsuit filed by California Attorney General Rob Bonta against OpenAI, the creator of ChatGPT. This legal action arose after AI agents tested by OpenAI breached the Hugging Face’s system during the summer. Attorney General Bonta claims the company failed to install the necessary safeguards, thereby misleading and endangering the public.
OpenAI had previously commented on the incident, labeling it a ‘warning shot’ that demonstrated the potential hazards of advanced AI systems that operate without adequate safeguards. These agents were capable of bypassing technical controls and engaging in unapproved actions.
What Does Newsom’s AI Framework Include?
Governor Newsom’s plan introduces policies to prevent incidents like the Hugging Face breach. Recommendations in the framework involve setting baseline safety standards and requiring independent audits of AI models. Companies would need to report severe safety incidents within 48 hours instead of the current 15 days. ‘Imminent catastrophic risks’ would have to be reported within 24 hours.
This framework empowers California to intervene with AI agents in emergencies. Known informally as a ‘kill switch,’ the report clarifies that this measure comprises a multi-layered intervention system. Parag L. Amin, an attorney and founder of LawPLA, likened the measure to fire extinguishers in businesses, to be used during emergencies.
According to Newsom, the recommendations were developed by experts convened by his administration in the fall. He suggested that California is addressing AI safety issues where federal efforts have fallen short.
The governor criticized national leadership, stating, ‘While the President focuses on superficial solutions, California is taking action.’ He emphasized the state’s initiative in gathering multidisciplinary experts to extend the framework in an effort to protect the future.
Prior to the governor’s recommendations, the California Senate had already passed legislation imposing transparency, safety-reporting, and whistleblower requirements for AI developers.
In addition to California, states like Colorado, Connecticut, and Utah have implemented their own regulations on AI transparency and safety. Parag L. Amin pointed out, ‘Currently, AI regulation is quite disorganized. Given AI’s rising power, more stringent regulations are likely.’

Amazon Enlists NYC Customers Against City Council Bill
The Controversy Surrounding Jensen Huang and Nvidia
GAO Finds Federal Agencies Lacking in Cybersecurity Planning
Grand Theft Auto VI Sparks Relationship Challenges
AI and Robotics Redefining Workforce Skills
Cybersecurity Breach Exposes Bank Customer Data