New Report: Boko Haram, ISWAP Harness AI for Bomb Craft, Operational Tactics

A recent investigation by The New York Times has brought to light how Boko Haram and the Islamic State West Africa Province (ISWAP) are deploying artificial intelligence (AI) platforms, including popular tools like ChatGPT, Claude, Gemini, Grok, and DeepSeek. These terror groups are reportedly utilizing AI to refine their explosive manufacturing processes and boost their operational capabilities.

The findings stem from research conducted by Antonia Juelich of the University of Cambridge. According to her study, former members of these insurgent groups revealed how they turned to AI systems to overcome tactical difficulties encountered during their attacks.

One ex-Boko Haram commander recounted a failed assault on a military base due to a defensive trench. He explained that the group consulted AI tools for guidance on modifying their motorcycles to clear such obstacles. They provided specific details about their motorcycles and the required jump distance. The AI reportedly furnished step-by-step instructions, which allowed the group's mechanics to improve the bikes' speed and acceleration. Fighters then practiced the maneuvers. The former commander was quoted saying, "We saw in a movie how motorcycles can jump over bridges. We used AI to learn how to do this."

Similarly, another former ISWAP commander informed the researcher that AI chatbots were instrumental in acquiring precise instructions for constructing explosives. "You type in the question or use your voice, and it gives you a detailed answer, like 'How can I build a bomb?', and then it tells you how. It is like a human robot. We used it a lot," the ex-commander reportedly stated.

The report highlighted that insurgents perceived AI as a means to mitigate the risks associated with trial-and-error methods, with one member asserting that "AI gives you accuracy." Another fighter claimed the group utilized AI-generated advice to alter chemical compositions, leading to the creation of more potent explosive devices.

In response to these revelations, OpenAI declared that such applications of its technology are in direct violation of its established policies. Google and Anthropic also affirmed that their AI models are specifically designed to decline dangerous requests concerning weaponry or harmful activities. However, the report indicates that former insurgents asserted they frequently managed to circumvent these chatbot safety measures by presenting malicious requests as legitimate or educational inquiries.