The resignation post that made Jacob Coxon one of the most talked-about names in AI this week is now facing a very different kind of scrutiny — not over whether his warnings about superintelligence are correct, but over whether the post itself was organic at all.
Coxon’s thread, in which he said he’d left Anthropic after three years working on pretraining at both the company and OpenAI, racked up an extraordinary amount of engagement in a short window. Within 24 hours, one tracker cited by tech commentator Theo Jaffee put the numbers at over 123 million views, 656,000 likes and more than 190,000 new followers on an account that had barely posted before. Elon Musk responded to that tally by saying he didn’t believe a new account with almost no prior activity had ever pulled numbers like that.
Connections To Newspeak House and AI-Safety Outlets
That scale of reach is what triggered a wave of scrutiny from several accounts that track influence operations and nonprofit funding networks.
Investigative researcher Parker Thayer, who works for the Capital Research Center, laid out a timeline he says looks less like an organic viral moment and more like a coordinated launch. Thayer noted that a Wall Street Journal exclusive quoting Coxon on his resignation reportedly went up only 18 minutes before Coxon’s own post did, which he argued points to advance planning with a media outlet rather than a spontaneous decision to speak out. He also pointed out that the first three accounts to quote-tweet Coxon’s thread — within roughly 15 minutes of it going live — belonged to Nathan Calvin of Encode AI, Peter Wildeford of the AI Policy Network, and Daniel Kokotajlo of the AI Futures Project, three organizations focused on AI policy advocacy. Thayer traced funding for those groups back to the Survival and Flourishing Fund, which he said has directed millions of dollars in grants from Skype co-founder and Anthropic investor Jaan Tallinn. He additionally flagged that Coxon received a Long-Term Future Fund scholarship from Good Ventures, the philanthropic vehicle of Dustin Moskovitz, in 2022, and that a program officer at Moskovitz’s Coefficient Giving was among the earliest accounts to amplify the post.
Separately, commentator Brian Chau published a thread connecting Coxon to Newspeak House, a London-based fellowship program for civic and political technologists, on the basis of a matching name, nationality and photo that Chau said appeared in the group’s published cohort materials alongside an FT photo credited to Coxon. Chau’s thread went on to describe Newspeak House’s funding history, which he said includes grants from the Long-Term Future Fund and the EA Infrastructure Fund, and argued that its past projects reflect a left-leaning political bent. He also noted, separately, Jordan Schachtel’s claim that Coxon had only been at Anthropic for about six weeks (other sources say 3 months) before resigning and that his social media accounts appeared to have been created shortly before the post went up. Newspeak House has also confirmed that Coxon was a “fellow”.
None of this amounts to confirmed proof of coordination, and it’s worth being clear about what is and isn’t established. The Survival and Flourishing Fund, the Long-Term Future Fund and Good Ventures are real, publicly documented funders in the AI safety and effective altruism space, and their grants to groups like Encode AI and the AI Futures Project have been previously disclosed by those organizations themselves. What Thayer and Chau are asserting is that shared funding sources and fast amplification timing point to a deliberate campaign — an inference that Coxon, Anthropic, and the named advocacy groups have not addressed on the record. Quote-tweeting a viral post within minutes is also simply what might happen on X when something is spreading fast, particularly among people who already work in a related field and follow each other.
The controversy is unfolding just as Anthropic’s Alignment Science lead Evan Hubinger publicly backed Coxon’s core claim that people inside frontier labs privately fear catastrophic outcomes from AI far more than their public statements suggest — a response that has nothing to do with the funding questions being raised, but complicates any narrative that dismisses the resignation as pure astroturfing. There have also been questions raised whether all the AI narratives on the internet are organic, with science YouTuber Sabine Hossenfelder saying that she was offered money by organizations to scare people about the impacts of AI.
Neither Coxon, Anthropic, nor the advocacy organizations named in Thayer’s and Chau’s threads have publicly responded to the claims as of this writing.