Overview
- Governor Gavin Newsom issued an executive order on Friday that speeds up California’s rollout of two recently passed laws and directs an expert panel to study an emergency “kill switch” for frontier AI systems.
- The order accelerates implementation of measures requiring independent third-party verification of high-risk models and asks for a near-term report on how a shutdown mechanism would work and be enforced.
- Security researchers have recently reported exploits that used advanced AI tools, including claims that a flaw in OpenAI’s public forum was taken over and that Anthropic software was used in internal probes, raising urgency for stronger controls.
- The move follows public safety warnings from figures inside the industry, such as Anthropic’s Dario Amodei, while other executives have pushed back against slowing development, exposing a sharp split over how to balance safety and innovation.
- With federal policy stalled and more than 600 state AI bills pending, California’s action could set practical rules for audits and emergency responses, shape what other states adopt, and affect how people’s jobs and online safety are managed.