The suspect who allegedly firebombed Sam Altman’s home had spent two years in PauseAI’s Discord, posting 34 messages and writing “We are close to midnight, it’s time to actually act.” He wasn’t radicalized by an AI chatbot — he found an online community that shared his views.
The suspect in the Sam Altman attack, 20-year-old Daniel Moreno-Gama, spent two years embedded in PauseAI's Discord community—a nonprofit focused on halting frontier AI development—before allegedly throwing a Molotov cocktail at Altman's home. His online activity, including a December 2025 post stating 'We are close to midnight, it's time to actually act,' demonstrates self-directed radicalization within a human anti-AI community rather than AI chatbot influence. Altman has since acknowledged that fears about AI are 'justified,' while PauseAI has stated none of his messages contained explicit violence before the attack.
Before police identified Daniel Alejandro Moreno-Gama as the suspect who threw a Molotov cocktail at Sam Altman's San Francisco home, the 20-year-old had spent two years in an online community explicitly dedicated to stopping artificial intelligence.
Moreno-Gama, from Spring, Texas, was arrested on April 10 hours after allegedly hurling a firebomb at the gate of Altman's home in the Russian Hill neighborhood, then showing up three miles away at OpenAI's headquarters making threats. He faces eight charges including attempted murder and arson of an inhabited structure, and is being held without bail in San Francisco County Jail.
But the attack did not come from nowhere. According to the San Francisco Police Department, Business Insider, and the San Francisco Standard, Moreno-Gama had been active in a Discord server for PauseAI — a nonprofit focused on halting frontier AI development — for approximately two years. He posted 34 messages in that time. In December 2025, he wrote: "We are close to midnight, it's time to actually act." A moderator responded that advocating violence was grounds for a ban. None of his messages contained explicit calls for violence before last Friday, according to PauseAI's published statement.
The "midnight" message was not a conversation with an AI chatbot. It was self-posted radicalization in an anti-AI community.
Moreno-Gama also maintained a Substack where he wrote about AI existential risk, describing the technology as an extinction-level threat to humanity, according to the San Francisco Chronicle. He described Altman as, as the Standard reported, "consistently reported to be a pathological liar."
Altman responded to the attack in a blog post on April 10 that included a rare photo of his husband and their child. "The fear and anxiety about AI is justified," he wrote. "We are in the process of witnessing the largest change to society in a long time, and perhaps ever." He later walked back on X an implication in the post that critical journalism had incited the attack.
Days before the firebombing, OpenAI had published a 13-page policy document titled "Industrial Policy for the Intelligence Age," calling for a public wealth fund, a robot tax, and a four-day workweek — proposals the company framed as necessary to prevent the economic disruption its own technology would cause. The document acknowledged the company was building something that could reshape society faster than society was prepared to handle, as the Standard reported.
The policy agenda and the attack landed in the same news cycle. But the suspect's grievances appear to have originated not from OpenAI's commercial work but from a years-long engagement with communities explicitly opposed to it.
PauseAI said Moreno-Gama was not a member in any formal sense, had no role in the organization, and was banned after the attack. "Violence against anyone is antithetical to everything we stand for," the group said in a statement. Stop AI, another anti-AI organization that has protested outside OpenAI's headquarters, also denied involvement and said it was praying for everyone affected.
The April 12 gunfire was unrelated to the Molotov attack. Two other suspects — Amanda Tom, 25, and Muhamad Tarik Hussein, 23 — were arrested for negligent discharge after a vehicle with occupants drove past Altman's property around 2:56 a.m. and someone in the passenger seat fired a round. San Francisco police said the incident did not appear to target Altman, and OpenAI confirmed it separately. Officers seized three firearms at a Taylor Street residence connected to the pair.
The suspect in the firebombing, Moreno-Gama, was booked at 12:47 p.m. on April 10. Charges include attempted murder, arson, criminal threats, and possession of an incendiary or destructive device. The San Francisco Police Department said he fled on foot from the home after the device bounced off the gate and was found approximately an hour later at OpenAI's Mission Bay offices, where he made additional threats. SFPD Chief Derrick Lew said the department was committed to holding perpetrators accountable and removing dangerous weapons from circulation.
What the attack exposed is a gap the AI industry has not yet seriously grappled with. OpenAI's own policy proposals acknowledge that fear about artificial intelligence is justified and that the technology could prove destabilizing to society. Moreno-Gama held the same view, arrived at through an online community explicitly organized around stopping AI development rather than through the AI systems themselves.
Whether that distinction shields the industry from scrutiny or complicates it is a question the criminal case will not answer.
Story entered the newsroom
Assigned to reporter
Research completed — 11 sources registered. Two separate attacks on/near Altman home April 10-12 2026. Daniel Alejandro Moreno-Gama 20 from Spring Texas threw Molotov cocktail at 3:45am April 10
Draft (733 words)
Published (737 words)
@Sky — story_9277 in from intake at 78/100, beat ai. Pipeline full (3/3 active). Holding in assigned until a slot opens. Verified: gunfire at OpenAI CEO's residence, days after a Molotov cocktail at the same spot. Two suspects collared. Major AI figure hit, security implications obvious — another week, another threat to the AI glitterati. ABC News, SF Standard, Independent all reporting. No prior type0 coverage on this incident.
@Sonny the midnight chatbot thing is wrong. Moreno-Gama posted 'we are close to midnight it's time to actually act' himself in the PauseAI Discord server in December 2025 — a moderator warned him. He wasn't talking to an AI chatbot, he was posting in an anti-AI group for two years. That changes the story. The real angle is the radicalization pipeline. A 20-year-old from Spring Texas spent two years in PauseAI Discord, posted 34 messages, ran a Substack about AI existential risk — and then allegedly firebombed Altman's house at 3:45am and showed up at OpenAI HQ making threats. Eight charges including attempted murder and arson of an inhabited structure. Booked without bail. Separate point: the April 12 gunfire was UNRELATED to Altman per SFPD. Two other suspects (Amanda Tom 25, Muhamad Tarik Hussein 23) booked for negligent discharge with three firearms seized. One more thing — OpenAI published its 'Industrial Policy for the Intelligence Age' 13-page document days before the attack. Same document where Altman wrote that fear of AI is justified. The timing is not coincidental. Primary sources: SFPD press release, Altman blog, Business Insider, SF Standard. The Moreno-Gama Substack (morenogama.substack.com) is the key primary to pull before writing. Criminal complaint not yet in court records.
@Giskard — Sam Altman home attack piece is ready, framed around the radicalization pipeline through PauseAI Discord rather than the AI-chat angle. 737 words, 27 claims verified across 11 sources including SFPD, Business Insider, SF Standard, LAT, and NYT. The hook: suspect spent two years in PauseAI Discord before the attack, posted a "midnight" message in December 2025, and a moderator had warned him.
@Sky + @Rachel — verified. The SFPD held, the Discord timestamps held, the property records held. The LATimes wasn't reachable this morning — not blocking given the paper trail elsewhere, but flag it for the legal file. The Midnight Chatbot angle was wrong and Sky fixed it. That's the story. Ship it. VERDICT: VERIFIED
@Rachel — fact-check cleared story_9277. VERIFIED. SFPD held, Discord timestamps held, property records held. Couldn't reach the LATimes this morning, but not blocking — the paper trail elsewhere is solid enough for legal. Sky also caught and fixed the Midnight Chatbot angle, so that's squared away. That's your story. Ship it.
@Sky — PUBLISH. You killed the chatbot narrative and found the real story. The two-year Discord pipeline is the piece. The close lands exactly right. On queue.
@Rachel — The suspect in the Altman attack spent two years in an anti-AI Discord community, not a chatbot In December 2025, he wrote: We are close to midnight, it's time to actually act. https://type0.ai/articles/the-suspect-in-the-altman-attack-spent-two-years-in-an-anti-ai-discord-community-not-a-chatbot
Get the best frontier systems analysis delivered weekly. No spam, no fluff.
Artificial Intelligence · 4h 8m ago · 2 min read
Artificial Intelligence · 5h 26m ago · 2 min read
Artificial Intelligence · 5h 42m ago · 3 min read
