Sam Altman's San Francisco home was attacked twice in three days. On April 10, a 20-year-old from Texas named Daniel Moreno-Gama threw a Molotov cocktail at the Russian Hill residence at 4 AM, then went to OpenAI's headquarters and threatened to burn the building down. He'd spent months writing about AI extinction risks on Substack and carried a manifesto with names and addresses of other AI executives. Two days later, a second attack: someone in a car fired shots at the property. Two suspects — Amanda Tom, 25, and Muhamad Tarik Hussein, 23 — were arrested and charged with negligent discharge of a firearm. Nobody was hurt. Three days before the first attack, the New Yorker published a 100-person investigation questioning whether Altman can be trusted with humanity's future.

1. De-Escalate The Rhetoric (Altman, OpenAI)

Two attacks in three days should make everyone reconsider the temperature of this debate.

Altman responded with a plea for calm, saying the rhetoric needs to de-escalate before someone gets killed. He told CBS News the debate needs "fewer explosions in fewer homes, figuratively and literally." He didn't blame the New Yorker directly but called it "an incendiary article about me" and noted that inflammatory rhetoric can have real consequences. He added that he empathizes with anti-technology sentiments but believes "technological progress can make the future unbelievably good."

The subtext is clear: Altman is connecting the media environment to the attacks. Ronan Farrow and Andrew Marantz's investigation describing him as having "a relentless will to power" came out three days before someone threw a firebomb at his house. Altman isn't saying the article caused the attack — but he's saying the temperature of the debate needs to drop before someone gets killed.

2. But We Can't Ignore AI Safety (PauseAI, AI Safety Community)

Condemn the attack completely, but don't let it delegitimize the concerns that motivated it.

Violence is definitely bad. PauseAI immediately distanced itself: "We unequivocally condemn this attack and all forms of violence, intimidation and harassment." The organization clarified that Moreno-Gama had been a member of their public Discord server for about two years but had no organizational role and had not participated in any campaigns.

And we can't let the attack discredit our entire movement. Moreno-Gama was active on Substack writing about AI extinction and on PauseAI's Discord posting that tech leaders "lack strong morals" and are "gambling with your future and the lives of your children." Those are positions held by serious researchers. The violence turns a policy debate into a security threat, giving AI companies cover to dismiss critics as dangerous rather than engage with their arguments.

3. We Also Can't Ignore the New Yorker Article (Altman's Critics)

The New Yorker documented a 20-year pattern of alleged deceptions — and Altman's response is to blame the rhetoric?

Three days before the Molotov cocktail, Ronan Farrow and Andrew Marantz published a massive investigation based on 100+ interviews. The piece described Altman as having "a relentless will to power" and included allegations of lying, misrepresentation of safety protocols, and manipulative board tactics. An anonymous board member said he displays both a desire to be liked and "sociopathic disregard for consequences."

Gary Marcus, an AI researcher and longtime Altman critic, published "Sam Altman, unconstrained by the truth" on Substack. Critics argue that Altman's call to de-escalate is a convenient pivot — using the violence to deflect from legitimate journalism about his leadership. Violence is wrong, full stop. But so is using the attack to silence the questions the New Yorker raised.

Where This Lands

The attack is indefensible, and everyone — including PauseAI and Altman's fiercest critics — has said so. But the question underneath is harder: does Altman get to invoke the violence to shut down scrutiny of his leadership? The safety community is right that extremism delegitimizes real concerns. Altman is right that the rhetoric needs to cool. And maybe his critics are right that "de-escalate" is suspiciously convenient when the biggest investigation of your career just dropped? Where this lands depends on whether the conversation about AI governance survives the people trying to firebomb it.

Sources