Boko Haram Used AI to Plan Motorcycle Attacks, Research Finds

A new research paper reveals that the Nigerian militant group Boko Haram has used artificial intelligence chatbots to plan and execute attacks, including modifying motorcycles for tactical advantages. The findings, based on interviews with former members, indicate that terrorist groups are increasingly turning to AI for battlefield assistance, despite safeguards implemented by AI companies.

The research, conducted by Antonia Juelich of Cambridge University and shared with The New York Times, documents an incident where Boko Haram fighters used a chatbot to learn how to modify motorcycles to jump over defensive trenches. The group had initially been thwarted by a trench surrounding a military base in eastern Nigeria. Using AI-generated instructions, mechanics altered the motorcycles for faster acceleration and top speed, and riders practiced jumps, sometimes with fatal outcomes, until they succeeded in a subsequent attack.

Juelich conducted nearly 60 interviews with 27 former Boko Haram members in Nigeria over the past year. Her field research found that terrorists also used chatbots to design explosives, fix weapons, and brainstorm attack strategies. The report states that large-language models have been consulted at every stage of military activity — mission preparation, operations, and post-mission analysis — contrasting with the commonly held perception that AI use by such groups is primarily for propaganda.

This evolution highlights a broader challenge for the AI industry. Chatbots have built-in limitations to prevent users from soliciting harmful information, but researchers have repeatedly found that individuals can circumvent these protocols by slowly coaxing models into revealing restricted knowledge. The findings come amid rising concerns about advanced AI models, which CIA Director John Ratcliffe recently likened to 'digital nuclear weapons.'

The Trump administration has pushed leading AI labs to allow government vetting of new models before public release, focusing mainly on cybersecurity risks. However, terrorism experts argue that the potential for AI to aid terrorist activities is underacknowledged. 'The terrorists are not waiting for us to make AI safe,' Juelich said, adding that their use of AI has been 'significantly underestimated in both scope and character.'

Daniel Byman, a terrorism expert at Georgetown University and co-author of a separate report on AI and terrorism, said groups are 'mixing and matching' from different AI systems to achieve their goals. The study underscores the need for stronger guardrails and oversight to prevent misuse of AI technologies by non-state actors.