Cambridge study finds Boko Haram and ISWAP use AI chatbots for bomb-making and tactics

A University of Cambridge research team found that jihadist groups Boko Haram and Islamic State’s West Africa Province (ISWAP) have embedded mainstream AI chatbots into bomb-making, tactics and operational security more systematically than previously known.

The finding comes from a report by Dr. Antonia Juelich for the Cambridge Programme on AI Science & Policy (CASP), based on face-to-face interviews with 27 former Boko Haram and ISWAP members conducted across northeast Nigeria in 2025 and 2026. Both factions have adopted the tools faster and more extensively than earlier analysis suggested, spanning weapons support, tactical planning, operational security and post-attack review, the report said.

Former members named ChatGPT, Claude, Gemini, Grok, Meta AI and DeepSeek among the systems used to design explosive devices, identify captured military hardware, troubleshoot weapons and assess battlefield tactics. In one account, AI guidance helped fighters modify motorcycles to jump defensive trenches after a failed base assault; a former ISWAP commander said the tools helped optimize explosive formulations for more powerful blasts.

According to the report, both groups institutionalized AI use through specialist units and internal training, and Islamic State-linked foreign operatives delivered in-person AI training.

The account rests on former insurgents’ testimony rather than direct technical observation, and the interviewees said they circumvented safety guardrails by disguising dangerous requests as legitimate projects. OpenAI, Google and Anthropic acknowledged the misuse risk, with Google and Anthropic saying their systems include protective guardrails.

The report lands as frontier AI vendors face growing pressure to show their safety filters hold up against determined real-world adversaries, not only lab red-teams.