TellWorldAI World Archive

Home›News›Policy

PolicyGenerative AI for everyone

California releases recommendations to strengthen its AI safety framework

On 18 November 2026 Governor Gavin Newsom released recommendations to strengthen California's AI safety framework, including state-set baseline standards, mandatory independent testing and state power to order emergency shutdowns.

Background

On 18 November 2026 Governor Gavin Newsom released new recommendations to strengthen California's AI safety framework [1]. The recommendations were developed under an executive order that convened a group of experts to recommend updates to the state's AI safety laws [1].

What happened

The recommendations propose baseline safety standards that the state, rather than companies themselves, would set for all AI companies [1]. They also propose mandatory independent safety testing and verification through all stages of model training, development and deployment [1].

Key details

The recommendations propose expanded safety incident reporting and verified risk-containment protocols, with state authority to order an immediate shutdown in an emergency [1]. They also propose giving the state power to cap the development and use of recursive self-improvement [1]. Recursive self-improvement refers to an AI system improving its own design or capabilities.

The release says the state is fast-tracking two bills. SB 813 sets up a framework for independent verification organizations, and AB 1405 creates a state registry for AI auditors [1].

Several terms in the recommendations need explaining. An executive order is a directive from a head of government that guides how the executive branch acts, and here it was used to convene experts for advice on updating state law. A baseline standard is a minimum requirement that applies to everyone. Independent verification means checking a system's safety claims with a party outside the company that built it. An AI auditor examines AI systems for compliance and risk, and a state registry is an official list of such auditors.

Impact

Newsom called on Congress and President Trump to review and adopt the state's framework, or to use it as a floor rather than a ceiling [1]. A floor sets a minimum standard that others can go beyond, while a ceiling sets a maximum.

Sources

  1. 1

Cite this page: California releases recommendations to strengthen its AI safety framework. TellWorld, https://www.tellworld.net/news/california-ai-safety-recommendations-2026. Updated 2026-11-20

Related news

View all