Tillbaka till Nyheter
PolitikAI Understanding genomgång

Newsom appoints experts to advance AI kill switch and oversight

Governor Newsom announced a group of leading experts to advise on his AI executive order, focusing on independent oversight and the creation of a potential 'kill switch' for rogue frontier models.

4 min readRead the linked source
Source-provided image accompanying Newsom appoints experts to advance AI kill switch and oversight
KällhänvisningKälla inspelad
Förläggare
gov.ca.gov
Källlänk
gov.ca.govhttps://www.gov.ca.gov/2026/09/23/governor-newsom-announces-world-leading-experts-to-deliver-on-his-ai-executive-order-including-advancing-creation-of-a-kill-switch/
Källtyp
Länkad källa – status för primär källa har inte fastställts.
SammanhangFörstå detta på 60 sekunder

Börja här

Nyckeltermer

Artificiell intelligens (AI)
Det breda fältet för att bygga system som utför uppgifter som kräver mönsterigenkänning, resonemang, språk eller beslutsfattande.
AI-justering
Arbetet med att få AI-system att bete sig enligt mänskliga intentioner, värderingar och säkerhetsbegränsningar.
AI säkerhet
Ett område fokuserat på att minska skadligt beteende, misslyckanden och missbruksrisker i AI-system.
Testa dig självAI Etik Quiz

Vad hände

Governor Gavin Newsom announced a panel of experts to advise on California's AI executive order, which aims to accelerate independent oversight and develop a 'kill switch' for frontier AI models.

Governor Gavin Newsom announced a group of leading experts to advise on recommendations required by his recent AI executive order. The order aims to accelerate independent oversight of AI companies and advance the creation of a potential 'kill switch' for rogue frontier models. The expert group, comprising individuals from technology, policy, and governance, will engage with technical researchers to guide California's and security governance.

The proposals under consideration include requiring independent third parties to be embedded in frontier AI companies to verify safety frameworks and independently assess safety evaluations. Additionally, the order directs the Government Operations Agency to accelerate implementation timelines for Senate Bill 813 and Assembly Bill 1405, which establish frameworks for independent verification organizations and a state registry for AI auditors, respectively.

The announced experts include Jason Goldman, a former White House Chief Digital Officer; Gillian Hadfield, a professor at Johns Hopkins University focusing on ; Alondra Nelson, a professor at the Institute for Advanced Study; and Rob Reich, a professor at Stanford University. These individuals will help apply expertise to the rapidly advancing capabilities of AI and address concerns regarding safety and security.

Källinformation: gov.ca.gov

Varför det spelar roll

This move establishes a concrete state-level mechanism for independent verification of claims and emergency shutdown capabilities, addressing a significant gap in current regulatory frameworks where federal oversight is absent or limited. It signals a shift toward mandatory third-party auditing and technical safety controls for high-risk AI systems.

California is positioning itself as a leader in AI regulation by creating a structured approach to independent verification and emergency shutdown capabilities. This is significant because no federal law currently requires AI companies to report dangerous incidents or mandates specific safety controls for frontier models.

The introduction of a 'kill switch' concept for frontier models represents a practical technical control that could mitigate risks associated with autonomous or rogue AI behavior. By embedding independent third parties in AI companies, the state aims to ensure that safety claims are verified by entities with no financial stake in the company's success.

This executive order builds on existing California legislation, including SB 53 (Transparency in Frontier Artificial Intelligence Act) and SB 813, creating a comprehensive regulatory framework that covers frontier-model safety, independent oversight, and child safety standards. It provides a model for other states and potentially the federal government to follow.

Interactive Mechanism

Interaktiv mekanism: hur det faktiskt fungerar

Utforska den underliggande tekniken bakom denna utveckling interaktivt.

Agent Lifecycle Stage:
1
User Intent & Planning: "Audit customer refund request #4092 and settle payment."
2
Tool Calling: Emits structured JSON call crm_get_transaction(id='4092').
3
Guardrail & Verification:🛡️ Paused: High-value action requires human operator sign-off.
4
Final Settlement: Refund recorded, email receipt dispatched, and audit log stored.
Core takeaway: An AI agent is not just a language model—it is a closed loop of planning, tool invocation, and environment feedback. Production systems require self-healing retries and strict human approval guardrails.
Interaktiv konceptkontroll+10 Points
AI Ethics Quiz

Impossibility results in algorithmic fairness (e.g. Kleinberg et al., Chouldechova) show what?

Vad du ska titta på härnäst

Watch for the specific technical recommendations regarding the 'kill switch' implementation, the timeline for SB 813 and AB 1405 implementation, and any federal response to California's expanding AI regulatory framework.

The specific technical specifications and feasibility of the proposed 'kill switch' for frontier models will be a key area of focus. How this mechanism would be implemented across different AI architectures and companies remains to be determined.

The timeline for the implementation of SB 813 and AB 1405 is critical. Accelerating these timelines could lead to earlier deployment of independent verification organizations and AI auditor registries, which would have immediate practical impacts on how AI companies operate in California.

The reaction from the AI industry and the federal government to California's expanding regulatory framework will be important. The Governor has called on Congress and the President to adopt or use California's framework as a floor for national standards, which could lead to further policy developments or conflicts.

Relaterade guider och frågesporter

AI-etikAI:s framtidAI-modeller förklarasTesta vad du vet – prova ett gratis AI-quizSlå upp en AI-term i vår ordlista
Hittade du detta användbart?