The attack on Sam Altman's San Francisco home came with a second strike two days later, and the OpenAI CEO is pointing a finger at his biggest rival.
Daniel Moreno-Gama, 20, allegedly threw a Molotov cocktail at Altman's $27 million Russian Hill residence around 3:45 a.m. on April 10, according to the federal criminal complaint.
No one was hurt (the firebomb bounced off the building). But less than two hours later, Moreno-Gama showed up at OpenAI's Mission Bay headquarters with a jug of kerosene, trying to smash through entrance doors with a chair before security confronted him.
He told security he planned to burn the building down and kill anyone inside.
San Francisco police arrested Moreno-Gama at the scene. Officers found incendiary devices, kerosene, and an anti-AI manifesto split into three sections, including one titled "Your Last Warning" that listed AI CEOs, board members, and investors as targets. The FBI said Moreno-Gama traveled from Texas to California with the intent of killing Altman.
Then on April 12, police arrested Amanda Tom, 25, and Muhamad Tarik Hussein after surveillance footage showed a passenger firing a gun from a Honda sedan outside Altman's home around 1:40 a.m., according to a San Francisco Police Department report. Officers found three firearms during a search of their residence.
Neither has been charged with a crime.
Altman addressed both attacks in an interview with podcaster Ashlee Vance posted Tuesday on Vance's "Core Memory" podcast. He directly blamed Anthropic for escalating tensions.
"I think the doomerism talk hasn't helped," Altman said. "I think the way certain other labs talk about us hasn't helped.
I think the way Anthropic talks about OpenAI doesn't help." The OpenAI-Anthropic rivalry has intensified for months. Anthropic was founded by seven former OpenAI employees who believed Altman wasn't prioritizing safety. In December, Anthropic CEO Dario Amodei mocked companies that declared "code reds," widely seen as a jab at OpenAI. Before the Super Bowl, Anthropic released an ad campaign ridiculing ads in AI chatbots (timed as OpenAI prepared to add ads to ChatGPT).
Altman fired back: "Anthropic serves an expensive product to rich people." The Information reported that Amodei wrote in a private memo calling OpenAI's messaging "safety theatre." "Sam is trying to undermine our position while appearing to support it," Amodei wrote.
Altman told Vance that after initial shock he went through "a real depressive cycle." On April 18 he published a blog post calling for de-escalation in AI debates and sharing a photo of his infant daughter. "Images have power," he wrote.
"Normally we try to be pretty private, but in this case I am sharing a photo in the hopes that it might dissuade the next person from throwing a Molotov cocktail at our house."
Moreno-Gama faces state and federal charges including attempted murder and attempted arson, carrying up to life in prison if convicted. US attorney Craig Missakian said prosecutors would treat it as domestic terrorism if evidence shows he acted to coerce policy change.
Moreno-Gama had posted online under the name "Butlerian Jihadist" (a Dune reference) suggesting "Luigi-ing some tech CEOs," an allusion to Luigi Mangione on trial for killing UnitedHealthcare's CEO. He joined Discord servers for PauseAI and Stop AI but was mostly inactive; Stop AI said he asked whether speaking about violence would get him banned and was told yes.
His parents said he had been suffering from a mental health crisis and had no prior criminal record. Diamond Ward, his public defender, called it overcharged: "This case is clearly overcharged.
This case is a property crime, at best."
Moreno-Gama's arraignment is set for May 5.
