A 20-year-old Texas man threw a Molotov cocktail at OpenAI CEO Sam Altman’s home in San Francisco’s Russian Hill neighborhood early Friday morning, then threatened to burn down the company’s headquarters an hour later, according to police.
Daniel Alejandro Moreno-Gama of Spring, Texas hurled the homemade firebomb at Altman’s gate at 3:45 a.m. April 10. The bottle with burning cloth bounced off the wall, and security extinguished the flames within minutes. No one was injured.
At 5:07 a.m., Moreno-Gama appeared at OpenAI’s Third Street headquarters in Mission Bay carrying a container he claimed held kerosene, police said. He threatened to set the building on fire. Officers recognized him from the earlier incident description and arrested him on the spot.
Moreno-Gama remains in custody on suspicion of attempted murder, arson and possession of an explosive device. The district attorney’s office has not yet filed charges.
AI Apocalypse Fears
Moreno-Gama had no personal connection to Altman. His own online posts reveal his attack was motivated by extreme fear of artificial intelligence, according to investigators.
He operated online under the pseudonym “Butlerian Jihadist” — a reference to Frank Herbert’s Dune universe where humanity destroys thinking machines. Communications researcher Nirit Weiss-Blatt, who preserved screenshots of his posts before account deletions, found Moreno-Gama participated in PauseAI’s Discord server. The group advocates halting development of advanced AI models.
In December 2025, he wrote on the server: “We are close to midnight, time to act.” A moderator immediately warned him that any calls for violence would result in a ban.
Moreno-Gama also published six lengthy essays on Substack from January to March 2026, according to SF Standard and The Decoder. In “A Eulogy for Man,” he argued AI development would lead to human extinction. He called technology leaders sociopaths and described Altman as a “pathological liar playing with humanity’s fate.” His writings cited Elon Musk and AI researcher Eliezer Yudkowsky, who has repeatedly warned that creating superintelligence could kill all humans on Earth.
Weiss-Blatt said Moreno-Gama’s case shows how online environments focused on catastrophic scenarios can accelerate radicalization of people already suffering from anxiety or depression, even when the organizations themselves don’t advocate violence.
PauseAI distanced itself from the attacker. The organization said Moreno-Gama joined their public Discord server about two years ago, left 34 messages and had no other involvement with the group. They blocked his account after the attack. A moderator began deleting his messages but stopped — they might be needed for the investigation.
Altman’s Response
Hours after the attack, Altman published a blog post showing photos of his family — his husband and son — for the first time. He wrote that he hoped this would stop the next person who decides to throw a Molotov cocktail at his home.
Altman acknowledged underestimating the power of words and narratives. He mentioned an “inflammatory article” about him published days before the attack. He said criticism of the AI industry often stems from sincere concerns that are well-founded, but called for de-escalating the rhetoric.
OpenAI confirmed it is cooperating with investigators. The company is currently hiring a head of industrial security in San Francisco.
Part of Broader Pattern
The attack on Altman’s home occurred four days after another incident targeting a politician over technology infrastructure. On April 6, an unknown assailant fired 13 bullets into Indianapolis city councilman Ron Gibson’s front door. A note under the doormat read “No Data Centers.”
Gibson had recently supported a $500 million Metrobloks data center in the Martindale-Brightwood neighborhood. Neither Gibson nor his 8-year-old son were injured. The FBI joined the investigation.
The two incidents within a week differ in character — one involving a radicalized individual with a manifesto, the other anonymous intimidation-style shooting. But both reflect growing resistance to technology expansion. Data centers face nationwide protests over energy and water consumption. Half of American adults fear AI, according to Pew Research.
The December 2024 murder of UnitedHealthcare CEO Brian Thompson in New York marked a turning point for corporate security. Companies sharply increased spending on executive protection in 2024, according to Reuters.
Security expert Kent Moyer, CEO of The World Protection Group, told SF Standard that Altman’s home security responded correctly. But all of Altman’s addresses — in Napa, Hawaii and San Francisco — remain easily findable online. OpenAI is already seeking a head of industrial security, while executive addresses remain publicly available on the internet.