Members of the Boko Haram insurgent group are increasingly turning to artificial intelligence (AI) tools, including ChatGPT, Gemini, Claude, Grok, and DeepSeek, to enhance their operations, according to a report by The New York Times, citing research conducted by Antonia Juelich, a researcher at the University of Cambridge.
The study, which was based on interviews with former Boko Haram fighters and commanders, found that insurgents had used AI-powered chatbots to seek information related to battlefield strategy, equipment maintenance, and other operational challenges.
One former Boko Haram commander recalled how the group sought AI-generated guidance after a planned attack on a military base failed because of a defensive trench surrounding the facility.
According to the former commander, the fighters provided details about the type of motorcycles they used and the distance they needed to cover, after which the AI chatbot suggested practical steps that could help improve the motorcycles’ performance. The group later modified the bikes for greater speed and acceleration before carrying out practice runs.
The research also found that some insurgents used AI systems to obtain technical information on repairing weapons and constructing improvised explosive devices (IEDs). Former members interviewed for the study described AI as a valuable resource that reduced uncertainty during planning, with one ex-commander reportedly likening the technology to “a human robot.”
According to the report, some fighters attempted to bypass the safety safeguards built into AI chatbots by disguising their requests as harmless educational or filmmaking projects. They also compared responses from different AI platforms to identify which systems disclosed more technical information.
Reacting to the findings, OpenAI said any attempt to use its AI products to facilitate terrorism or violent activities is a clear violation of its usage policies. The company added that it continues to strengthen its safety systems to detect and prevent misuse.
Similarly, Google and Anthropic said their AI models are designed to refuse requests that could facilitate harmful or illegal activities and that they are continually improving safeguards to reduce the risk of abuse.
The findings have renewed concerns among security experts about the growing misuse of generative AI by extremist groups and criminal organizations, highlighting the need for stronger safeguards, international cooperation, and continuous monitoring as artificial intelligence becomes more accessible worldwide.