On April 11, 2026, 20-year-old Daniel Moreno-Gama threw a Molotov cocktail at Sam Altman's house at 3:45 AM, then walked three miles to OpenAI headquarters threatening to burn it down. He was booked on suspicion of attempted murder. This wasn't random. Moreno-Gama was an active PauseAI member with six community roles. His Discord handle was "Butlerian Jihadist." His online presence was filled with doomer content. Four months before the attack, he'd recommended Eliezer Yudkowsky and Nate Soares' book "If Anyone Builds It, Everyone Dies" to his followers. On his own Substack, he estimated the probability of AI-caused extinction as "nearly certain" and wrote, "We must deal with the threat first and ask questions later."
Alexander Campbell, writing on his Substack Campbell Ramble, argues this was the logical endpoint. When you build a framework on certainty and keep escalating the rhetoric, cheap talk eventually meets reality. He points to the Center for AI Safety's national spokesperson saying on camera that the "correct response is to walk to the labs across the country and burn them down." A PauseAI leader said an Anthropic researcher "deserves whatever is coming to her." When someone flagged this rhetoric in PauseAI's Discord, the mods deleted the post. Campbell notes that when Yudkowsky was asked why he wasn't attacking data centers himself, his answer wasn't "because violence is wrong." It was "because it wouldn't work yet."
The restraint is strategic, not moral.
Not everyone accepts Campbell's framing. Zvi Mowshowitz condemned all political violence as both immoral and ineffective, emphasizing that most major AI safety voices have consistently warned against it. Critics point out that people with mental illness often gravitate toward political causes without representing the movement, and that the vast majority of people who believe in existential risks don't resort to violence. Fair points. But Campbell's observation holds: when you tell people with certainty that everyone they love will die, and that builders are doing it anyway, someone's eventually going to act. PauseAI deleted Moreno-Gama's Discord messages. They can't delete the syllogism that put them there.