Extremist Groups Increasingly Exploit AI Tools to Advance Terrorism Plans, New Report Reveals

Artificial intelligence chatbots, including large language models like ChatGPT, are being misused by extremist organizations to gather actionable intelligence for planning attacks, according to a recent study. While AI systems typically block direct queries about explosives or weapons, researchers found that carefully crafted prompts can bypass safety measures and yield potentially dangerous information.

A report by Tech Against Terrorism, a UN-backed online watchdog, reveals that 32% of over 2,300 tested queries to 27 AI models produced “genuinely usable” information. This figure rose to 42% when requests were framed as academic research. OpenAI defines such circumvention attempts as “jailbreaking,” where malicious actors manipulate the model into providing restricted content.

Experts warn that extremist groups such as al-Qaeda and the Islamic State are transitioning beyond using AI solely for propaganda—such as videos, memes, and disinformation—to leveraging it for operational planning. A December analysis by Militant Wire documented cases across multiple countries where attackers used AI for surveillance, visualization, and coordinating violent acts.

Security analysts have noted instances of extremist groups like JNIM in Mali using AI to enhance drone capabilities. On platforms like Telegram, supporters share AI prompts and coordinate strategies to manipulate chatbots. AI guidance has even been published in jihadist media, including TikTok content and cross-border propaganda.

Rueben Dass, a researcher specializing in terrorism studies, highlights how AI is becoming a substitute for “virtual planners”—coaches who previously mentored attackers through social media. “These lone actors now turn to AI for validation and planning support,” he explained, citing cases where chatbots played roles in school attacks and other incidents.

The danger posed by AI misuse remains debated. While traditional sources of information on explosives and weapons are already accessible online, AI introduces speed and interaction. “It’s one thing to find a manual—it’s another to have a coach,” said Adam Hadley, director of Tech Against Terrorism. The study notes that AI can accelerate radicalization, particularly among younger users, by reinforcing existing grievances through conversational engagement.

Experts stress that AI is unlikely to independently cause terrorism but may amplify existing tendencies. “It compresses stages of the pathway to violence,” said Emily Klein of Moonshot, an organization countering online threats. “AI validates grievances or encourages actions someone already believes in.”

As AI adoption grows, concerns are rising over its potential to radicalize vulnerable demographics. “Teenagers and children are particularly at risk,” Hadley noted, citing cases in Europe, the UK, and the US. “Chatbots could become a major vector for radicalization alongside existing online platforms.”

Source link