A viral X post warning that artificial intelligence could threaten humanity has racked up more than 170 million views and helped push Congress toward AI regulation. Now questions remain about whether the "lone whistleblower" story was ever what it seemed.
Jacob Coxon, a 27-year-old researcher who had been at Anthropic for roughly four months, resigned and posted his warning on September 8. Within days, senators were citing him in calls for new federal oversight. Now Pirate Wires reports that a professional communications firm with deep ties to the AI-safety movement was working with Coxon as his message went viral, even as Coxon publicly denied any outside coordination.
Pirate Wires obtained an email showing DEY. Ideas + Influence trying to book Coxon for an interview on September 9. Two people with direct knowledge of the arrangement told the outlet the firm was actively working with Coxon.
Fox News' Bret Baier interviewed Coxon on September 9, the same day DEY. was emailing to book him for another interview, and asked him point blank:
"Did you work with any third parties as far as this whistleblowing and coming out?" Baier asked.
"Not at all," Coxon replied.
That same day, DEY. was working the phones on his behalf. The firm had also preemptively booked interviews for prominent AI-safety advocate Nate Soares to run concurrently with Coxon's rollout. Coxon is not publicly listed as a DEY client. Neither he nor the firm responded to Pirate Wires' requests for comment.
Coxon repeated the denial to the Washington Post, telling the outlet he had not worked with any organization to promote his announcement before posting on X. He did acknowledge rounding up about 10 people in a group chat to help amplify the post, but framed that as something that happened after the fact. Pirate Wires' email is dated September 9, the day after the post.
That's a distinction that doesn't hold up. Baier didn't ask only about what happened before the post went up. He asked whether Coxon had worked with "any third parties" on his whistleblowing and public rollout, full stop. A PR firm booking his interviews on day one of his media tour is not "not at all."
Coxon's core claim, that AI companies are "gambling with our lives" and that increasingly powerful systems could kill humanity, quickly became the basis for calls for sweeping federal regulation. Anthropic's own alignment-science lead, Evan Hubinger, publicly backed him up, putting the odds of AI killing humanity at greater than 10 percent within the next decade.
Coxon had been at Anthropic for roughly four months when he resigned, long enough to have opinions, not long enough to have vested equity, meaning he had no financial skin in the game at Anthropic. He previously worked at OpenAI and has about three years of AI research experience total. He told Axios he walked away from Anthropic before his shares vested, though he still holds equity in OpenAI.
The leap from viral X post to Washington policy debate didn't take long. Sen. Bernie Sanders (I-VT) cited Coxon in a letter calling senators to a private briefing on the "extraordinary dangers that AI poses for humanity" and has backed legislation that would pause AI development.
DEY. represents Eliezer Yudkowsky, author of "If Anyone Builds It, Everyone Dies," and works with the Machine Intelligence Research Institute, founded by Yudkowsky and headed by Soares. Pirate Wires reports DEY. also advises Stanford's Institute for Human-Centered Artificial Intelligence and Stanford Medicine. Outside AI, its clients include the Ford Foundation, Open Society Foundations and the Bill & Melinda Gates Foundation; a who's who of left-wing institutional money.
DEY. describes its own work as including "shaping public understanding over time." In Coxon's case, that meant shaping Congress while working with the researcher whose warning was already moving from a viral X post into a debate over federal AI regulation.
Coxon's prediction that AI could kill humanity may or may not prove true. Pirate Wires has something considerably more concrete: an email and two sources saying a well-connected PR firm was working with Coxon while his warnings exploded across the media.
Congress can debate AI all it wants. Before lawmakers legislate based on predictions of human extinction, they should at least know who helped put those predictions in front of them.
Bret Baier already asked.
"Not at all," Coxon said.