Anti-AI Violence Is Spreading. Here's What's

In the early hours of April 10th, someone threw a Molotov cocktail at the San Francisco home of Sam Altman, CEO of OpenAI. The bottle bounced off the house. Nobody was hurt. The attacker then walked to OpenAI's headquarters and tried to smash through the glass doors with a chair, telling security personnel he'd come to burn the building down and kill anyone inside. The U.S. Department of Justice identified him as a 20-year-old from Texas who had written an anti-AI manifesto arguing that human extinction was imminent — and that AI executives were the ones to blame.

It would be tempting to file this under lone wolf, disturbed individual, isolated incident. Except it isn't isolated.

In the same month, CBS News reported that Indianapolis City Councilor Ron Gibson had 13 shots fired at his home, with a note left behind reading "No Data Centers" — this after Gibson had publicly supported a data center development in his district. In May, two teenagers carried out a deadly attack in San Diego, leaving behind a manifesto laced with anti-technology ideology: complaints about AI-generated content flooding the internet, fears of surveillance, calls for sabotage of the industrial complex. In Rome, an Italian influencer was arrested and charged with organizing a series of anti-tech attacks. In France, a group attempted to torch machinery inside an AI data center, citing the industry's ties to military technology, border enforcement, and the war in Gaza.

And then there's Berlin.

In September 2025, an anarchist group set fire to two power transmission masts, triggering a blackout that knocked out electricity for tens of thousands of households and businesses — some for more than 24 hours. The group framed it as a deliberate "system interruption" and explicitly called on others to replicate the tactic, arguing that research, science, and technology are "entangled with war, environmental destruction, and social control." Separately, around the same time, a different group torched a cable bridge in the same city, affecting over 45,000 homes and businesses for several days. That group described its actions as "emergency self-defense" against the dominance of technology.

Law enforcement has noticed the pattern. Wired reports having obtained documents from a New York intelligence and counterterrorism bureau warning of "widespread upheaval in response to AI adoption in the coming years."

Physicist and science communicator Sabine Hossenfelder covered these incidents in a recent video, and her framing is worth examining both for what it gets right and where it strains. She's correct that these aren't random acts. She's also correct that the landscape is genuinely ideologically varied — the attackers share an anti-tech sensibility, but their specific grievances span extinction risk, anti-capitalism, anti-surveillance, anti-militarism, and straightforward NIMBY frustration at data centers landing in residential neighborhoods. As Hossenfelder notes: "These are small groups that often have other motives such as opposing one or the other, war, capitalism, or generally anyone they believe has power over anything."

That ideological sprawl is actually important context. This isn't a coherent movement with a unified theory — at least not yet. What these incidents share is a target (AI and its physical infrastructure) and a conviction that conventional opposition isn't working. Petitions didn't stop the data centers. Local zoning appeals didn't stop the data centers. Viral social media outrage definitely didn't stop the data centers. Violence, for these actors, reads as the logical escalation of a grievance that institutional channels have declined to address.

Whether that logic is sound is a separate question. The Berlin blackouts didn't slow AI development. They inconvenienced tens of thousands of ordinary people who have nothing to do with tech policy — a fact that tends to undermine rather than advance a political argument.

The scale of public opposition to AI, though, is harder to dismiss. Gallup reported in May 2026 that seven in ten Americans oppose AI data centers being built in their local area, with nearly half saying they're strongly opposed. That's not a fringe position — that's a majority, and a motivated one.

Hossenfelder's point here is structurally sound: "Most opposition to AI and data centers is not violent, but if you have upset that many people, it's not surprising that some of them react with violence." That's not an endorsement of violence; it's a statistical observation. When you generate that volume of genuine public grievance — over noise, energy consumption, water usage, property values, and the broader cultural anxieties that AI has come to represent — a small fraction of that opposition will eventually express itself through means that go beyond letters to elected officials.

The question Hossenfelder raises, and which law enforcement is already acting on, is whether these scattered incidents will eventually coalesce into something more organized. Her speculation: they probably will, and likely internationally. The historical precedent for this kind of diffuse ideological violence organizing into coherent movements — the radical environmentalism of the 1990s and 2000s being an obvious comparison — suggests she's not wrong to worry about the trajectory.

Where Hossenfelder's analysis gets philosophically interesting — and where her conclusions deserve some scrutiny — is at the end of her video, where she pivots from six minutes of documented public failure and escalating opposition to declare that "more intelligence is the best thing that can happen to our planet."

It's a striking move. She's spent the bulk of her time cataloguing incidents that are, in large part, downstream of broken governance: communities that feel steamrolled, democratic channels that have failed to register opposition, and a technology deployment that's moving faster than any meaningful public consent process could hope to keep up with. And then she concludes that the answer is more of what's generating the friction.

That's not necessarily wrong — you can believe both that AI development is net positive and that the current rollout is badly managed. But Hossenfelder doesn't quite make that argument; she just asserts the optimistic conclusion and moves on. A physicist cataloguing systemic failure and then endorsing the system's accelerant isn't being inconsistent so much as incurious. The evidence she's assembled points toward a governance problem. "More intelligence" doesn't automatically solve a governance problem — sometimes it sharpens it.

What the Berlin blackouts, the Indianapolis shooting, the San Diego attack, and the Molotov cocktail in San Francisco have in common is that none of them changed anything about how AI is being developed or deployed. The companies are still building. The data centers are still going up. The venture capital is still flowing.

What they've done, instead, is generate exactly the kind of coverage that will make AI companies argue for more security, more regulatory protection, and more deference from local communities — which is to say, they've handed the industry a political gift.

The activists who genuinely believe they're slowing AI down by attacking power masts are, at minimum, catastrophically miscalculating. Whether they're miscalculating about AI's dangers too is the question that most deserves a real answer — because if the next incident is larger, the time for that conversation will already be gone.

By Marcus Chen-Ramirez, Senior Technology Correspondent, Buzzrag