Will AI Concerns Become a New Source of Terrorism?

CSIS has posed a question in its title — Will AI Concerns Become a New Source of Terrorism? — and then, in the manner of all Washington think-tank products, answered it with a dataset and a hedge. Drawing on its U.S. Terrorism Incidents Dataset, the analysis concludes that AI-related grievances could become a recurring source of terrorism, while conceding the evidence does not yet show that they have. Two attacks — that's the corpus. Somewhere in an office with good chairs and a JSTOR subscription, two data points have been promoted to a 'phenomenon' and given a research question.

Those two attacks, we are told, establish that serious violence is possible. One would like to congratulate the analysts on this finding, which is also true of cricket, planning permission and the post office queue. The same paragraph concedes they do not show a broader campaign has arrived — the sort of sentence that only exists because someone in communications asked for a caveat that could survive being quoted. The phrase 'AI-related grievances' is doing magnificent load-bearing work here, holding up a category wide enough to contain anything from a chatbot's feelings to an actual ideology with a manifesto and a stationery budget.

Still, credit where it's due: this is the rare AI-threat paper that resists the urge to project a red line onto a chart and call it a forecast. The pattern is nonetheless familiar — name a threat category early, publish the dataset, and wait. If a third incident arrives, the paper is vindicated; if none does, the paper was prudent — either way, the citation stands. Think of it as an actuarial table for civilisational dread, filed under 'recurring,' which is the word you use when you want to have already been right. The question mark in the title is the most honest thing in it.