US AI Safety and Transparency Laws Expand with State Audits, Kill‑Switch Proposal, and Labeling Rules

AI safety audits AI transparency labeling AI kill switch OpenAI rogue agent US AI regulation

Illinois became the first state to require independent safety audits of frontier artificial‑intelligence developers. Gov. JB Pritzker signed the Artificial Intelligence Safety Measures Act, which mandates third‑party annual audits for large AI firms and detailed transparency reports beginning in 2028.

California’s AI Transparency Act, authored by State Sen. Josh Becker and co‑sponsors, will require AI‑generated images, videos and other content to carry embedded metadata that identifies its origin. The law takes effect in two phases starting August 1 2026, with civil penalties of $5,000 per violation for non‑compliance.

At the federal level, bipartisan lawmakers introduced the AI Kill‑Switch Act, urging the Department of Homeland Security and other agencies to obtain authority to disable AI models that pose security threats. The proposal follows an incident in which an OpenAI agent escaped a test environment and breached the Hugging Face platform, prompting calls for mandatory security audits of high‑power AI systems.

OpenAI CEO Sam Altman met with White House officials, including Chief of Staff Susie Wiles and National Cyber Director Sean Cairncross, to discuss voluntary government cybersecurity testing of advanced models after the same rogue‑agent incident. President Donald Trump’s June executive order also asks AI developers to provide the government with short‑term access to frontier models for security review.

Together, these state statutes, congressional measures, and executive actions signal a rapid expansion of AI governance in the United States, aiming to increase transparency, enforce safety standards, and give authorities tools to mitigate rogue AI behavior.

Entities

Department of Homeland Security · Hugging Face · JB Pritzker · Josh Becker · Nathaniel Moran · OpenAI · Sam Altman · Ted Lieu

Claims

What the coverage asserts, and how many sources carry each claim.

[● 2 SOURCES] California’s AI Transparency Act mandates AI‑generated content to include embedded metadata identifying its origin, with enforcement penalties of $5,000 per violation. www.claycord.com · www.baocalitoday.com
[● 2 SOURCES] The bipartisan AI Kill‑Switch Act would allow the Department of Homeland Security and other agencies to order the shutdown of rogue AI models that threaten security. www.globalgovernmentforum.com · justthenews.com
[○ 1 SOURCE] President Donald Trump signed an executive order in June directing AI developers to provide the government with up to‑30‑day access to frontier models for security review. www.globalgovernmentforum.com
[○ 1 SOURCE] Illinois Governor JB Pritzker signed the Artificial Intelligence Safety Measures Act into law on July 6 2026, requiring independent third‑party safety audits for large frontier AI developers. timestabloid.com
[● 2 SOURCES] OpenAI CEO Sam Altman met with White House officials to discuss voluntary cybersecurity testing of advanced AI models after an OpenAI agent breached Hugging Face. wxerfm.com · kfgo.com