Daniel Moreno-Gama was arrested for allegedly attempting to murder OpenAI CEO Sam Altman and burn down OpenAI's headquarters.
SAN FRANCISCO -- Months before his arrest for allegedly attempting to murder the chief executive of OpenAI, Daniel Moreno-Gama suggested "Luigi'ing some tech CEOs" in an internet chat.
The Texas college student casually referenced Luigi Mangione, the accused UnitedHealthcare CEO killer, during an online conversation with producers of "The Last Invention" podcast, according to screenshots shared with The Wall Street Journal. They wanted to interview him for a series on artificial intelligence.
In January, Moreno-Gama recorded the interview, detailing his path from curious internet nerd to a crusader obsessed with AI's dangers. He said those chilling words shouldn't be taken literally, according to a recording shared with the Journal.
"I understand the frustration with a person who might advocate for that, but it's not practical," Moreno-Gama said. "It's not worth it."
Last week, authorities alleged Moreno-Gama, 20, traveled from the Houston area to San Francisco, threw a Molotov cocktail at Sam Altman's mansion and then attacked OpenAI's headquarters' entrance, planning to burn the building down.
The incident spotlights a brewing anticorporate fervor in some internet subcultures, amplified by the national attention on Mangione, an Ivy League-educated 27-year-old who has pleaded not guilty. Only weeks ago, Chamel Abdulkarim, 29, was charged with sparking a blaze at a Kimberly-Clark warehouse in Southern California. He complained about wages and compared himself to Mangione, federal prosecutors said.
"A lot of people are going to understand," Abdulkarim said, likening what he did to when "Luigi popped that m -- -- ," according to the federal complaint. Abdulkarim has pleaded not guilty.
In the OpenAI case, investigators found a manifesto, allegedly Moreno-Gama's, warning that AI would wipe out humanity. It included a message to Altman: "If by some miracle you live, then I would take this as a sign from the divine to redeem yourself..."
Moreno-Gama faces federal and state charges, including attempted murder and arson. He has yet to enter a plea. Diamond Ward, his public defender in the state case, said prosecutors overcharged Moreno-Gama, calling the incident a "property crime, at best."
"It is unfair and unjust for the San Francisco District Attorney and the federal government to fearmonger and exploit this young man's vulnerability simply due to the high-profile status of the people involved," Ward said.
Ward said Moreno-Gama has a history of autism and mental-health issues, adding that the actions "appear to have been driven by an acute mental-health crisis, not a desire to harm."
Moreno-Gama's parents said they have been trying to get him mental-health treatment and were concerned about his well-being. "He is a very caring person and has never been arrested before," they said in a statement." Until very recently, he was working hard at a restaurant and attending college classes."
The attack prompted Altman last week to share a rare photo of his family on his blog, saying it "might dissuade the next person from throwing a Molotov cocktail at our house, no matter what they think about me."
In a statement, Altman acknowledged concern about artificial intelligence -- but said that shouldn't lead to violence. "We should de-escalate the rhetoric and tactics," he wrote.
News Corp, owner of The Wall Street Journal, has a content-licensing partnership with OpenAI.
In the podcast interview, released in edited form Wednesday by media startup Longview, Moreno-Gama recalled thinking OpenAI's ChatGPT was "awesome" during his high school years, because he could "cheat on everything."
Online, Moreno-Gama used the handle Butlerian Jihadist, referencing a fictional war between humans and thinking machines from the sci-fi classic "Dune." Though he recorded the podcast interview under the moniker "Discord Dan," the podcasters decided to reveal his identity in light of recent events. The Journal was able to independently confirm his identity.
Andy Mills, Longview's editor in chief, said the podcast initially offered anonymity to Moreno-Gama, but that "his own actions and online statements have since established a clear link between his pseudonym and his real identity."
The podcasters, Mills said, had reached out to Moreno-Gama while working on an episode exploring perspectives of those who believe "we may need to consider violence to protect humanity from what they believe is a digital supermind that may lead to our extinction."
During the interview, Moreno-Gama described learning about AI's dark side by reading prominent AI critics like Eliezer Yudkowsky, who warned in a 2023 Time magazine opinion column that "the most likely result of building a superhumanly smart AI, under anything remotely like the current circumstances, is that literally everyone on Earth will die."
"I was like, OK, I hope he's kind of wrong, but over time, I realized very few of his main criticisms ever got refuted," Moreno-Gama said.
He debated strangers online, yet remained unconvinced by their arguments against Yudkowsky's thesis. He pestered his parents and friends about the risks of AI.
"I kind of became a bit, like, annoying, a bit autistic about that," he said.
Moreno-Gama's mother suggested that he find a group, and in 2024 he joined PauseAI, which advocates halting development of the most powerful AI systems. Impressed by the tireless activists he met online, he tried to emulate them, evangelizing and writing articles on Substack.
Maxime Fournes, Pause AI's CEO, said Moreno-Gama joined the group's public Discord server and posted 34 messages. "None of his messages contained explicit calls to violence," Fournes said. "We unequivocally condemn this attack and all forms of violence."
On another online forum called Stop AI, Moreno-Gama last year asked, "Will speaking about violence get me banned?" according to that group. He stopped posting after he was told "Yes."
"Stop AI has always adhered to nonviolent activism," the group said.
During the podcast interview, Moreno-Gama bristled at stereotypes of hysterical "AI doomers." If people read as much as he did about AI, he argued, they would be just as concerned as he was.
"Before we even think about violence, we need to exhaust all our peaceful means," he said. "I think protesting, I think sharing information, I think doing podcasts like this, that needs to come way before we even consider that."
But like many young people, Moreno-Gama viewed Mangione as a political touchpoint. He said he disagreed with the alleged killer's tactics, but he noted it was interesting how with Mangione, "a lot of people were able to excuse it."