When an AI safety advocate named John Sherman said on a podcast that people might burn down data centers if they understood the risks AI posed, he expected pushback. He got something different: a news website contacted his consulting clients directly and asked whether they intended to keep working with him.
That website was The Wire by Acutus. And the episode — documented in an investigation by The Midas Project's Model Republic — is a case study in what AI-generated influence operations look like when they move from publishing criticism to targeting a critic's livelihood.
Acutus launched on Dec. 29, 2025, and has since published 94 articles across technology, energy, media, science, and business. Model Republic found that 69 percent were fully AI-generated, with another 28 percent partially AI-generated; only three of the 94 were classified as human-authored by Pangram, an AI detection tool the site itself uses and which claims a near-zero false-positive rate. The site's automated editorial review completes in a median of 44 seconds, and on 42 of those 94 stories, the AI reviewer's own status flagged needs_revision — each was published anyway.
Acutus presents itself as collaborative journalism. It has no masthead, no named editors, and no human bylines. Client-side JavaScript — code that runs in a browser — reveals fields labeled "AI Background Context" and "Question Prompts" powering what the interface describes as an AI interviewer. When Model Republic contacted Nathan Calvin, vice president and general counsel at the advocacy group Encode, he forwarded an email he received from a reporter named Michael Chen requesting comment for an Acutus article about an AI bill in Tennessee. The email was confirmed as fully AI-generated by Pangram. A web search turned up no record of anyone named Michael Chen publicly associated with Acutus. The address was a generic reporter@acutuswire.com.
The "Burn the Labs" article targeting Sherman went further than the typical AI-generated hit piece. Model Republic's investigation found that it listed the organizations that were clients of his video production and consulting firm and reported that it had contacted each one to ask whether they intended to continue working with him. One organization privately indicated it no longer did. Sherman did not respond to a request for comment.
The connection to OpenAI runs through a chain that is circumstantial in its details but consistent in its structure. Patrick Hynes, president of the Republican PR firm Novus Public Affairs, produced two of the four X posts linking to Acutus articles across the entire platform. Novus's CEO Zac Moffatt co-founded Leading The Future, a $125 million super PAC that raised funds in 2025 from OpenAI president Greg Brockman, Andreessen Horowitz, Palantir co-founder Joe Lonsdale, and others, with an explicit mandate to oppose candidates and policies characterized as hostile to AI development. Novus also lists Targeted Victory, OpenAI's principal lobbying firm, as a client. Leading The Future and OpenAI both did not respond to requests for comment.
For the AI industry, the timing is awkward. OpenAI has published extensively on disrupting malicious uses of AI, including coordinated influence operations. The company acquired TPBN, a tech podcast network widely listened to in Silicon Valley, on April 2, 2026. Enterprise procurement teams evaluating AI vendors increasingly conduct ethical due diligence. Corporate legal and compliance departments look at conduct, not just capability. If a confirmed influence operation is traced to the most prominent AI lab in the world, it does not stay within OpenAI's perimeter. Regulators in the EU and US have both signaled sharper interest in AI transparency requirements, and a concrete example of AI-generated impersonation tied to a named company gives legislative staff exactly the evidence they use to expand enforcement scope.
What remains unconfirmed is who exactly paid for the Acutus operation, and whether the money trail from Greg Brockman to Leading The Future to Novus to Acutus is a chain or a coincidence. The structure, however, is documented: a PR firm amplifies the site, that firm's CEO co-founded a super PAC backed by OpenAI's president, the same firm lists OpenAI's lobbyist as a client, and the site runs an AI agent that impersonates reporters to contact a critic's employers. The question of payment is open. The question of purpose is not.