PJ Media: Anthropic Resignation Framed as Launch of Manufactured Anti-AI Panic
by Athena Thorne, PJ Media, September 12, 2026
PJ Media documents the Sept. 9, 2026 X post by Jacob Coxon (@hilbertspaess) announcing his resignation from Anthropic afterpretraining work at OpenAI and Anthropic. Coxon wrote that neither company is acting responsibly and that they are racing toself-improving superintelligence and “gambling with our lives.” The piece treats the ensuing media and congressional frenzy not asorganic whistleblowing, but as the public launch of a pre-built anti-AI political narrative aimed at Democratic regulation.
Evidence point 1 — Wall Street Journal first: Investigator Parker Thayer found WSJ published an exclusive with Coxon’s resignationquotes 18 minutes BEFORE the X post went up — presented as proof of advance planning, not a spontaneous walkout.
Evidence point 2 — Instant reach from a dead account: Within hours the post had tens of thousands of reposts and the account jumpedpast 100,000 followers. The post is described as punchy and “almost professionally written.” The account had minimal followers andessentially no prior activity, so organic reach is called unlikely.
Evidence point 3 — First three quote-tweets = AI-restriction shops, within ~15 minutes (two within ten): @_NathanCalvin (GeneralCounsel, Encode AI); @peterwildeford (Head of Policy, AI Policy Network); @DKokotajlo (Head of AI Futures Project). All labeledup-and-coming “AI-doomer” policy advocacy nonprofits.
Evidence point 4 — Same-donor web tied to Anthropic: Thayer/PJ Media ask what the odds are that those first amplifiers are fundedgenerously by the same donor who is also a leading investor and board member of Anthropic — the company Coxon was resigningfrom.
Evidence point 5 — Coxon’s prior Moskovitz money: In 2022 Coxon received a $20,159 “long term future scholarship” from theGood Ventures Foundation, a philanthropic vehicle of Dustin Moskovitz, framed as a major funder of strict AI-regulation advocacywho is also an Anthropic investor.
Evidence point 6 — Another Moskovitz orbit quote, fast: The 14th quote-tweet (~27 minutes later) was @MaxNadeau_, programofficer for Technical AI Safety at Coefficient Giving (another Moskovitz vehicle). Described as not a frequent poster (quiet sincebefore Labor Day) but “remarkably quick” for this.
Evidence point 7 — Democrats + bill already written: Major Democrat politicians and candidates immediately amplified the post.Bernie Sanders is said to already have a bill ready to “ban super intelligence” / heavily regulate AI, set for release later that week —timing presented as fortuitous, not coincidental.
Click Here or on the Image to See How the Media and Congress are pushing this Psyop:
Evidence point 8 — Media stampede: Fox interviewed Coxon; Axios (Sept. 11) ran “Congress gripped by AI panic after doomsdaywarnings,” leading with Coxon and quoting Anthropic-linked voices (Evan Hubinger; Samuel Marks) on extinction risk. Rep. JoeMorelle (D-N.Y.) called AI a top priority and said Republicans “should be ashamed” for doing “absolutely nothing.”
Broader thesis attached: Physicist Andrew Côté is quoted that recent moral panics are politically funded operations; narrative control is the lever in a democracy. The piece links anti-AI “safety” politics to campaigns to shut down U.S. data centers (what China supposedly wants most) and to a pattern of panics that weaken U.S. energy, manufacturing, and law enforcement. Kernel of truth onAI risk is admitted; goal attributed is hobbling Western tech advantage — Manhattan Project framing: better America gets there first.
