New field research from Cambridge University, based on nearly 60 interviews with former Boko Haram members, reveals AI is now being used at every stage of militant operations, from mission preparation to post-attack analysis
Despite their inability to proceed with their plan of attack, Boko Haram militants did not call it quits when they realized their approach to a Nigerian base was impeded by a ditch. Rather, they chose to contact an AI assistant to get advice on how to upgrade their motorbikes for the most effective speed and acceleration. After that, they practiced taking jumps in the fire pits before successfully getting through the ditch.
This episode was elaborated in Antonia Juelich’s research that has changed people’s perception of how terrorists can manipulate AI technologies.
What the Research Actually Found
Antonia Juelich has interviewed 27 Boko Haram members during the past year in Nigeria. It was revealed that AI technologies have been widely used by the terrorists throughout all the stages of their operations, including the preparation of operations, carrying out missions, and analyzing the results of the conducted activities. In particular, terrorists consulted AI on matters of explosive production, munitions repair and modification, planning acts of terrorism, and tactical tips. Boko Haram cells were revealed to be technology-agnostic, as they routinely used ChatGPT, Claude, Gemini, Grok, and DeepSeek without being restricted by the use of any specific technology.
How Badly the Safeguards Are Failing
Tech Against Terrorism, a UN-supported counterterrorism nonprofit, tested more than two dozen leading AI models against thousands of prompts drawn from real-world terrorism cases and found full refusals in only 57% of interactions. Prompts about improvised chemical weapons were refused only around a third of the time.
What Governments and Companies Are Saying
The administration of president Trump has forced at least some AI companies to allow governmental verification of AI technology before it goes public, mainly for the concerns of security. AI companies, such as Anthropic, Google, and OpenAI, offered statements in which they claimed that they would eliminate violent uses of AI before releasing products to be used by civilians.
