Ex-Anthropic researcher linked to PR firm amid AI fear debate

By Splitfeed EditorialLast updated Sep 26, 2026, 19:17 UTC

On September 25, 2026, reports from multiple outlets revealed that Jacob Coxon, a former Anthropic researcher who had publicly warned that AI could "kill us all," had worked with a public relations firm called DEY. The same firm has also worked with prominent AI skeptic Eliezer Yudkowsky. That same day, Joe Lonsdale, an investor in Anthropic, stated that AI companies are deliberately stoking fear about AI risks in order to shape policy in their favor. Anthropic itself has not been reported as having commented on either development.

The story matters because it sits at the intersection of two contested questions: whether AI doom warnings represent genuine expert concern or a coordinated messaging strategy, and whether the AI industry benefits from amplifying fear to entrench incumbents through regulation. The link between Coxon and a PR firm that also represents other prominent skeptics does not on its own establish coordination or bad faith, but it gives critics a concrete foothold for arguing that the safety discourse is being managed rather than spontaneous. Lonsdale's remarks broaden the accusation well beyond one researcher, implicating AI firms as a class in what he frames as deliberate fear-mongering for regulatory advantage.

The pro-AI camp, as reflected in Lonsdale's statement reported by Reuters, argues that safety rhetoric from within the industry is being weaponized to influence policymakers rather than to protect the public. The anti-AI camp reads the same facts in a different direction: for accounts like cookie crumbs and Hagen Blix on Bluesky, the PR firm connection is evidence of industry propaganda and regulatory capture, with suspicion directed at Anthropic specifically. Lauren Jones raises a related concern, pointing to Anthropic's financial support for NPR as a reason to distrust media coverage of AI safety. Hagen Blix describes the firms involved as "clearly industry propaganda." The middle ground camp has not weighed in on this story, though a typical middle ground position would likely focus on whether professional communications assistance inherently compromises a warning, and would call for transparency rather than treating the PR link as automatically disqualifying.

The clearest thing to watch is whether Coxon, DEY, or Anthropic respond publicly to the reports, and whether Lonsdale's remarks prompt any documented response from the companies he is implicitly criticizing. Any disclosure about the scope or funding of DEY's work with AI skeptics would significantly shift the weight of the argument in either direction.