A former Anthropic researcher’s public resignation has set off a fast-moving debate in Washington over how — or whether — the federal government should regulate artificial intelligence, after Sen. Bernie Sanders (I-VT) introduced legislation that would pause advanced AI development and impose steep criminal penalties on developers.
Jacob Coxon, 27, announced on Sept. 8 that he was leaving Anthropic after three years doing pretraining research at both OpenAI and Anthropic, two of the country’s leading AI labs. In a public thread, Coxon said neither company was acting responsibly, writing that they “are racing straight to self-improving superintelligence and gambling with our lives.”
In a subsequent interview with CNN’s Anderson Cooper, Coxon said advancing AI capabilities could surpass human intelligence by 2030 or sooner, and that systems acting with independent volition could “cause extreme havoc.”
“I think that if you extrapolate into the future the level of capabilities of these AIs, with the same independent volition, they could cause extreme havoc,” Coxon said.
Sanders responded by introducing the Ban Artificial Superintelligence Act, which would suspend advanced AI development until a federal regulator establishes safety rules. The bill has drawn attention not just for its aim but for its enforcement mechanism: it would allow prison sentences of up to 20 years for developers and researchers working on advanced AI systems.
The scope of that penalty has become a flashpoint among critics who otherwise might support some form of AI oversight. For a liberty-minded audience, the concern is straightforward — a law that criminalizes research with two-decade prison terms hands enormous discretionary power to whichever federal regulator is charged with defining what counts as “advanced AI,” and to whichever administration controls that regulator.
Questions About Motives on All Sides
Coxon’s resignation and the timing of the legislative response have also drawn skepticism from online commentators, some of whom have pointed out financial and political ties among figures who quickly amplified the story, questioning whether the episode was less a spontaneous whistleblower moment than a coordinated push for regulation. Others have speculated, without confirmation, that an undisclosed incident involving a frontier AI model may have prompted AI executives including Elon Musk, Dario Amodei and Sam Altman to signal support for slowing development. None of these claims have been independently verified.
Separately, former Secretary of State Hillary Clinton has called for what she described as an “Operation Warp Speed” for new AI controls, adding another prominent voice to the push for federal intervention.
The Underlying Tension
The debate underscores a genuine dilemma for anyone skeptical of both runaway AI development and runaway government power. On one hand, a researcher with direct experience inside two major labs is warning that safety is being sidelined in a race toward more powerful systems. On the other, the primary legislative response on offer pairs a development freeze with criminal penalties severe enough to invite concern about regulatory overreach — and about which companies might benefit if smaller competitors are priced out by compliance costs that incumbents can more easily absorb.
Sanders’ bill has not yet advanced beyond introduction. No hearing date or companion legislation has been announced.
Add comment