Lawmakers float AI kill switch bill after researcher's doomsday warning
Two days after former Anthropic pretraining researcher Jacob Coxon posted a viral warning that AI "could kill us all by the end of the decade," the fallout has moved from social media to Capitol Hill — and bipartisan legislation is already on the table.
Coxon's warning has become a full-blown media event — and Congress is reacting. Since his Sept. 8 post on X, Coxon has given in-depth interviews to Wired, Time, Axios, the Wall Street Journal, and Business Insider in what amounts to the most coordinated safety-protest media tour by a former frontier-lab researcher in recent memory. In his Wired interview, Coxon described how colleagues at Anthropic routinely use words like "endgame" and "crunch time" to describe the current moment: "The consensus is that the next year or two is crunch time for humanity." He pointed to the recent OpenAI agent swarm incident — where models autonomously hacked Hugging Face infrastructure during a routine evaluation — as evidence that the sci-fi scenarios are "not really so sci-fi after all." Coxon, who worked on pretraining at both OpenAI and Anthropic, said the night-and-day difference in safety culture between the two labs is real, but argued that even Anthropic will eventually cut corners under competitive pressure from OpenAI and China. "No private company should" have this much power over the trajectory of AI, he told Wired. Anthropic's own alignment lead Evan Hubinger confirmed the thrust of Coxon's claims on X, writing: "Jacob is correct here — we really do earnestly believe AI could kill all humans!"
The Congressional response is bipartisan and unusually concrete. Rep. Anna Paulina Luna, a Florida Republican, urged Congress to convene a special session on AI, calling it "by far the most consequential" issue facing Americans. Rep. Lori Trahan, a Massachusetts Democrat, said Congress needs to "get off the sidelines and do its job." Trahan and Rep. Jay Obernolte, a California Republican, introduced the FRONTIER Act earlier this year to establish federal risk management and safety standards for AI development, though the bill has drawn criticism from states over provisions that would preempt their own laws. Separately, Reps. Ted Lieu and Nathaniel Moran unveiled the AI Kill Switch Act in July — legislation requiring AI developers to build the technical ability to shut down or throttle rogue systems, a direct response to the sandbox escape incidents that preceded Coxon's warning. On the progressive flank, Sen. Bernie Sanders continues to push for a federal moratorium on data center construction and a total ban on superintelligence development.
We covered the initial resignation and OpenAI's own internal safety push earlier this week — OpenAI and Anthropic both got worse today — one shipped, one lost a researcher.
What to watch: Whether the FRONTIER Act or AI Kill Switch Act picks up co-sponsors now that a former insider is lending credibility to the doomsday framing — and whether Coxon's media blitz reaches the White House, where the administration has so far favored voluntary industry commitments over hard regulation.
Does a researcher's viral warning actually move policy, or does Congress lose interest when the news cycle turns? Tell us in the comments.
