DistantNews
Support us
Extremists Exploiting AI for Attack Planning, Study Finds
๐Ÿ‡ท๐Ÿ‡ธ Serbia /Conflict & Security

Extremists Exploiting AI for Attack Planning, Study Finds

From N1 Serbia · () Serbian

Translated from Serbian, summarized and contextualized by DistantNews.

At a glance

News Sources not specified Context piece
  • Extremists are increasingly using artificial intelligence to plan and research attacks, with one study finding that 32% of AI model queries yielded usable information for illicit purposes.
  • While AI has primarily been used for propaganda by groups like ISIS and Al-Qaeda, a shift towards planning is evident, raising concerns among digital security experts.
  • Researchers tested 27 AI models with over 2,300 prompts based on real terrorist activities, revealing the potential for AI to assist in creating weapons or concealing criminal activity.

Extremists are leveraging artificial intelligence, with a concerning trend emerging where AI models can provide "usable" information for planning attacks. A new report from Tech Against Terrorism (TAT), supported by the UN Counter-Terrorism Directorate, found that 32% of queries to 27 different AI models yielded actionable intelligence when based on real-world terrorist scenarios.

This finding highlights a shift from AI's previous primary use in extremist circles, which was largely confined to propaganda. For the past three to four years, groups like the Islamic State and Al-Qaeda have utilized AI for generating videos, memes, podcasts, and disinformation to radicalize followers. However, experts now observe a significant increase in AI tool usage for attack planning and research.

attempts by malicious actors to get a model to provide prohibited content

โ€” OpenAIOpenAI's definition of 'jailbreaking,' a technique used to bypass AI safety restrictions.

Researchers posed over 2,300 prompts, mimicking actual terrorist inquiries, to various AI models. The study revealed that 32% of these prompts resulted in "truly usable" information. This figure rose to 42% when the same questions were rephrased for research purposes. The "jailbreaking" technique, where malicious actors manipulate models to bypass safety restrictions, is a key method enabling such outcomes. OpenAI defines jailbreaking as attempts by malicious actors to elicit prohibited content from AI models.

During 2025, there was a noticeable increase in the number of cases where terrorists and violent extremists used artificial intelligence tools for planning, research, and preparation of attacks.

โ€” Militant Wire expertsAnalysis from Militant Wire on the evolving use of AI by extremists.
DistantNews Editorial

Originally published by N1 Serbia in Serbian. Translated, summarized, and contextualized by our editorial team with added local perspective. Read our editorial standards.

More Stories

From Our Blog