Will AI Concerns Become a New Source of Terrorism?

CSIS | Daniel Byman and Riley McCabe

Only two attacks, both in April 2026, count as terrorism motivated by opposition to artificial intelligence. On April 6, someone fired 13 rounds into the Indianapolis home of Councilor Ron Gibson, who backed a $500 million data center, and four days later Daniel Moreno-Gama, 20, attacked OpenAI chief executive Sam Altman’s San Francisco residence and the company’s headquarters.

Neither attack killed anyone, and no organisation has been tied to either. Vandalism, including more than 240 destroyed Flock Safety licence-plate cameras by early September 2026, falls outside the CSIS definition, which requires intent to kill. Motives run from fear of human extinction to local grievances over water and power, and white supremacist groups have borrowed the cause. The record is thin. Executives and local officials may prove easier targets than guarded facilities. The risk would grow if attackers copy one another, extremist communities reward violence, or opponents conclude that elections and protest cannot slow AI.

Comment

The three incidents raise a detection problem more than a motive problem, since the usual starting point is a network. Counterterrorism against organised movements, such as the post-9/11 effort against jihadist groups, relies on watching members talk to and fund one another. Daniel Moreno-Gama, Seth Aaron Pendley and Ethan Paul Early appear to have acted independently, so there was little such trail to follow. Moreno-Gama’s written call to kill AI executives, with names and addresses, suggests planning, yet the brief reports no agency seeing it before April 10.

Pendley was arrested in Fort Worth in April 2021 after taking what he believed was C-4 from an undercover FBI employee. Early’s thermite plan against xAI’s Memphis data center ended when friends persuaded him to abandon it; one of them then reported him to police. That tip route has limits: it works only if the plotter confides in someone first. In Pendley’s and Early’s cases, the break came from human contact, and neither route depended on watching a network.

Moreno-Gama would be the harder case on this reading. He appears to have acted independently, and the brief records nobody close to him raising an alarm before the attack on OpenAI’s headquarters.

Strategic Question for Discussion

What happens to the detection of plots like Ethan Paul Early’s against the xAI data center in Memphis if the next plotter confides in nobody?

My assessment is that such a plotter would be close to invisible to tools built for organised movements, because Early’s plan reached police only through a friend who chose to report. That points to local contacts, rather than collection, as the main early warning in cases of this kind. A plotter who tells no one would leave only purchases of materials, such as the ingredients Early had begun buying, as a signal.

Share your assessment in the comments below.

To read in details click here