Burn the Labs: How John Sherman’s AI Doomerism Collided with Real-World Violence

AI Risk Network and Guardrail Now's John Sherman’s violent rhetoric collides with real-world attacks, raising questions about accountability, influence, and silence from corporate partners

When John Sherman tweeted that artificial intelligence is “an intruder in your home” that is “here to kill you, your kids, your pets, and everyone you know and love,” it landed in a week when that kind of language no longer reads as abstract.

Hours earlier, a Molotov cocktail had been thrown at the San Francisco home of Sam Altman. In the days that followed, the picture grew sharper. The suspect in that attack has now been charged with attempted murder. Authorities say he maintained a list of AI executives across multiple companies described as a “kill list,” alongside online activity tied to anti-AI communities. In a separate incident, two additional suspects are under investigation after firing at Altman’s residence.

Across the country, in Indianapolis, a city councilman who supported a data center project had 13 shots fired at his home. A note was left behind: “NO DATA CENTERS.”

The debate over artificial intelligence has been escalating for months. What changed this week is that the language and the violence began to look like they were moving in the same direction.

Sherman sits near the center of that shift.

He is the founder of Guardrail Now, a small but increasingly visible advocacy effort focused on AI extinction risk. His message is simple and repeated often: artificial intelligence is not a distant concern. It is an immediate, existential threat.

His rhetoric is sharper than most.

Sherman has called for people to “walk to the labs across the country and burn them down. Like, literally." He has argued that humanity’s “best chance” may be a catastrophic “warning shot” that kills “a few million people.” He has accused AI executives of “attempted murder of ALL of us.”

The language is not theoretical. It names an enemy. It suggests a response.

In Washington, that tone is now drawing a reaction that cuts across party lines. A Republican policy staffer, granted anonymity to speak candidly, described Sherman’s rhetoric as crossing a line. “John Sherman and the fringe anti-AI crowd around Guardrail Now aren’t serious policymakers — they’re radicals whose language is now colliding with real-world violence. Calling for labs to be burned down and framing AI as an ‘intruder’ here to ‘kill your kids’ isn’t advocacy; it’s incitement. Any business, politician, or conservative who associates with Sherman or these groups is playing with fire and undermining American innovation and public safety. Stay clear of them. This is not who we are.”

A national Democrat who supports federal AI regulation offered a similar assessment. “Mainstream Democrats support responsible guardrails on AI to protect workers, privacy, and national security — but we will not allow radicals like John Sherman to hijack that position. His calls to ‘burn the labs’ and his grotesque talk of a ‘warning shot’ that kills millions have no place in our party or any legitimate policy conversation. When extremists start compiling ‘kill lists’ and violence follows, it’s time to draw a hard line.”

Sherman is not operating entirely on the margins. He was hired as Director of Public Engagement at the Center for AI Safety by Dan Hendrycks, a prominent figure in AI risk circles. His past statements were already public at the time. He was later dismissed after they drew scrutiny, but only after being placed in a role designed to shape public understanding of AI risk.

That tension — between fringe rhetoric and institutional proximity — runs through nearly every part of Sherman’s profile.

By day, he runs Storyfarm, a Baltimore-based video production and consulting firm. Its client list reads like a cross-section of corporate America and major institutions: the United States Agency for International Development, the University of Virginia, Match Group, Under Armour, Medtronic, SAP, Johns Hopkins University, MedStar Health, Starbucks, Ace Hardware, Uber, Mastercard, and McDonald's.

Many of those organizations rely on artificial intelligence in core parts of their business.

Storyfarm does too.

On its website, the company promotes its own use of AI tools and services, including a dedicated section highlighting how it integrates AI into client work. The same firm whose founder calls for AI labs to be destroyed markets itself as fluent in the technology.

The contradiction is not subtle.

Acutus contacted each of the organizations listed above to ask whether they were aware of Sherman’s statements and whether they intended to continue working with his firm. The companies did not respond to repeated requests for comment.

One organization indicated privately that it had not worked with Storyfarm for some time. Shortly after outreach, its logo was removed from the firm’s website.

Kathleen Kennedy Townsend, the former Lieutenant Governor of Maryland and now a research professor at Georgetown University, serves as a senior advisor to the group. She did not respond to repeated requests for comment and did not denounce the calls for violence.

The structure behind Guardrail Now raises additional questions.

Dads Against AGI Inc, doing business as Guardrail Now, is registered as a Baltimore-based 501(c)(3) nonprofit. It reports less than $50,000 in annual revenue to the IRS, qualifying it to file only a Form 990-N, according to IRS public records reviewed by Acutus. That filing provides no detail on finances, donors, or expenditures. At the same time, the organization lists eight team members on its website, a footprint that appears larger than its disclosed financial scale would suggest.

There is no public record of its donors.

What exists instead is a highly visible advocacy platform, operating with limited transparency, advancing some of the most aggressive rhetoric in the AI debate at a moment when that debate is still being defined.

Earlier this week, Acutus reported on the convergence of AI risk narratives, funding, and power. Sherman’s story sits inside that convergence. It shows how a loosely structured organization, amplified by institutional ties and left largely unchallenged by corporate partners, can help shape the tone of a policy fight before formal guardrails are in place.

His rhetoric does not argue for regulation. It argues that the threat is already here and that conventional responses are insufficient.

That framing is no longer contained to online discourse.

An alleged attacker charged with attempted murder. A reported list of AI executives identified as targets. Multiple incidents of violence tied to opposition to AI infrastructure.

There is no evidence that Sherman directed any of these acts. But the distance between the language and the targets has narrowed.

In the space between those two things — rhetoric and action — is where this story now sits.

And it is getting smaller.