Someone firebombed Sam Altman's home. Online, some cheered.

Around 4 a.m. on April 10, a 20-year-old from Spring, Texas, named Daniel Moreno-Gama allegedly threw a lit Molotov cocktail at the driveway gate of Sam Altman's San Francisco home. Nobody was hurt. Hours later, police arrested him outside OpenAI's headquarters, where he had thrown a chair against the glass doors trying to get in. In his bag: a document listing the names and home addresses of AI executives and investors, along with writings outlining anti-AI grievances.

The San Francisco District Attorney has since charged him with attempted murder.

What happened before the firebomb

Four days before the Altman attack, someone fired 13 shots at the front door of Indianapolis city councilman Ron Gibson's home. His 8-year-old son was inside. A note left at the scene read: "NO DATA CENTERS." Gibson had recently voted to approve a data center in his district. No one was injured, but the message was explicit.

These are not isolated incidents that happened to share a news cycle. They share a target profile, a grievance, and a cultural reference point. Moreno-Gama had posted in online anti-AI spaces about "Luigi-ing" tech CEOs, a reference to Luigi Mangione, the man accused of killing the UnitedHealthcare CEO in December 2024. Investigators confirmed those posts as part of the evidence trail.

The movement and the fringe

Pause AI and Stop AI, the two most established nonviolent anti-AI activist organizations, moved quickly. Both publicly disavowed the attack and stated that Moreno-Gama was not and had never been a formal member.

That's true. And it doesn't fully close the door.

In the weeks before the attack, Moreno-Gama posted 34 times on Pause AI's public Discord server. He separately contacted Stop AI to ask whether violent talk would get him banned. Told yes, he left. His documented radicalization happened in and around these spaces, even if those spaces never endorsed it. The organizations are not the movement's violent fringe. But the fringe is watching the same conversations.

This is the dynamic that's hard to hold in your head: a movement can be genuinely nonviolent at its core and still produce conditions where someone tips over the edge. That doesn't make the movement responsible for the attack. It does mean the movement has a problem it can no longer ignore.

The grievances that got here

The anger didn't start online. It started in places like artists' studios, freelance writers' inboxes, and small towns watching power infrastructure buckle under the weight of new data centers.

More than 1,500 AI-related bills are currently moving through state legislatures. Maine has passed a moratorium on data centers larger than 20 megawatts, pending a governor's signature. An Indianapolis councilman's home got shot up because he voted yes on a data center. These are not the actions of a movement that feels heard.

The people who oppose AI's expansion are not a monolith. Some are worried about existential risk. Some are worried about their livelihoods. Some are worried about what a 200-megawatt facility does to a rural power grid. All of those concerns have legitimate policy channels. Most of the people who hold them are using those channels. A small number are not.

The Luigi effect

What happened after the firebomb is its own story. On X and in corners of Reddit, posters framed Moreno-Gama as a folk hero, drawing direct comparisons to Mangione. One user called both men "heroes" in the same post. Multiple accounts called the attack "justified."

This is the Luigi Mangione effect in a different sector: an act of political violence gets aestheticized online before the suspect is even arraigned. The celebration is not coming from the organized anti-AI groups. It's coming from a diffuse, radicalized fringe that has absorbed the lesson that violence generates sympathy and attention, and that certain platforms will amplify that sympathy before anyone thinks too hard about it.

Moreno-Gama allegedly studied the template. He posted about it. He used the word. That's not a coincidence.

Why it matters

The industry's response so far has been largely internal. Silicon Valley executives are adding home security details. Prosecutors are treating this as attempted murder, not political protest. The internal debate has shifted, according to multiple reports, from messaging strategy to threat assessment.

But that's the smaller problem. The larger one is structural. Legitimate economic anxiety, the kind that drives 1,500 state bills and gets an artist to a Pause AI Discord server, is being laundered through an online culture that cheers violence and calls it resistance. The people doing the cheering are not the people who lost their commissions. They're feeding off the anger those people feel.

At the same time, an industry that spent years dismissing AI backlash as technophobia is now learning that the anger has an address. Not a metaphorical one. A literal one, at 4 a.m., with fire.

Turning the temperature down requires someone to actually want to. The question is who thinks that's their job.

When the people with the most power to reshape how AI is deployed keep treating opposition as a PR problem, and when online spaces keep converting grief into glorification of violence, who is actually responsible for what comes next?

Originally published as an Instagram carousel on @recul.ai.