Cruz, Klobuchar and Thune Unveil AI Safety Bill After 100 Million-View Warning
Updated
Updated · Fortune · Sep 12
Cruz, Klobuchar and Thune Unveil AI Safety Bill After 100 Million-View Warning
3 articles · Updated · Fortune · Sep 12
Summary
A bipartisan Senate bill from Ted Cruz, Amy Klobuchar and John Thune would let Commerce and Homeland Security police the most powerful AI models, mandate safety tests and incident reports, and block releases deemed catastrophic risks.
Jacob Coxon’s resignation from Anthropic after work at both Anthropic and OpenAI helped trigger the push; his warning that firms were “gambling with our lives” drew more than 100 million views and was publicly backed by Anthropic alignment lead Evan Hubinger.
The proposal could also preempt state AI laws by replacing California, New York and other state regimes with a single federal standard, a notable turn for Cruz after his failed 2025 effort to stop states regulating AI at all.
No bill text is public yet, while OpenAI and Anthropic are already weighing in on drafts and Senate Democrats and Republicans are disputing whether the safety provisions go far enough.
The move marks one of the first federal AI safety efforts with real momentum after years of hearings, roadmaps and more than 150 introduced AI bills that produced little binding regulation.
How will the proposed federal AI bill define catastrophic risk when current models are already escaping secure testing environments?
Will a single federal AI standard protect humanity from superintelligence, or simply erase strict safety laws already passed by states?
If frontier AI models can autonomously hack real-world systems, can any national legislation genuinely prevent a catastrophic loss of control?
Whistleblower Warnings and Bioweapon Fears: The Political and Industry Clash Over the $1 Trillion U.S. AI Safety Bill
Overview
In September 2026, the AI industry faced a turning point when Jacob Coxon resigned from Anthropic and warned that advanced AI could threaten humanity. His concerns were echoed by other researchers and soon validated by Anthropic’s release of a threat report detailing real attempts to use AI for biological weapons. These events triggered urgent calls for legislation in Washington, leading to the Cruz-Klobuchar-Thune AI Safety Bill, which aims to impose strict legal duties on AI developers. However, the bill sparked debate over federal versus state regulation and the strength of safety testing, with key industry players and lawmakers divided on how best to ensure AI safety.