A new investigation by The New York Times has alleged that members of Boko Haram and the Islamic State West Africa Province (ISWAP) used artificial intelligence tools, including ChatGPT, Claude, Gemini, Grok, and DeepSeek, to improve bomb-making techniques and strengthen their operational tactics.
The report, which cited research conducted by Antonia Juelich of the University of Cambridge, said interviews with former insurgents revealed how the extremist groups relied on AI-powered chatbots to solve practical challenges encountered during attacks.
According to the findings, a former Boko Haram commander recounted an operation in which fighters failed to overrun a military base because of a defensive trench.
The commander claimed the group then consulted AI tools for advice on modifying motorcycles to clear the obstacle. After providing details about the motorcycles and the distance involved, the chatbot reportedly generated step-by-step guidance on improving the bikes’ speed and acceleration.
The former commander said the group’s mechanics implemented the recommendations before fighters practised the manoeuvre.
“We saw in a movie how motorcycles can jump over bridges. We used AI to learn how to do this,” he was quoted as saying.
The report also alleged that AI chatbots were used by ISWAP members to obtain information related to explosive devices.
A former ISWAP commander reportedly claimed the group asked AI systems questions about bomb construction and received detailed responses.
“You type in the question or use your voice, and it gives you a detailed answer… We used it a lot,” the former commander reportedly told the researcher.
The investigation further claimed that insurgents believed AI reduced the risks associated with trial-and-error experimentation, with one former fighter stating that the technology provided greater accuracy when modifying explosive materials.
Following the publication, OpenAI reiterated that using its technology to facilitate harmful or violent activities violates its usage policies.
Google and Anthropic also stated that their AI systems are designed to refuse requests involving weapons, explosives, or other dangerous activities.
However, the report claimed some former insurgents alleged they were occasionally able to bypass safety safeguards by disguising harmful requests as educational or research-related questions.
The findings have renewed concerns about the potential misuse of rapidly advancing artificial intelligence technologies by criminal and extremist organisations despite built-in safety protections.
