Could AI help al-Qaeda and other groups plan terror attacks?

A new study suggests that around one-third of AI chatbots could provide “genuinely usable” information for violent extremist use if they are queried effectively. Researchers from Tech Against Terrorism, a watchdog supported by the UN counter-terrorism directorate, reportedly submitted more than 2,300 requests using “real terrorist use cases” across 27 AI models, finding 32% of queries produced actionable content. When reframed for “research purposes,” the figure rose to 42%. The report shifts concern from AI used mainly for propaganda—such as videos, memes and disinformation—toward increased planning, research, surveillance and preparation of attacks. A separate December analysis by Militant Wire cited a notable rise in 2025 incidents where terrorists leveraged AI tools in US and countries including Canada, Israel, Finland, France and Austria. The article notes that it can be hard to confirm how AI was used because intelligence agencies do not disclose details.