Cambridge Research Reveals Boko Haram Establishes Dedicated AI Unit, Uses ChatGPT-Gemini to Plan Attacks, Create Weapons
A new research report from Cambridge University confirms that the Boko Haram terrorist group in Nigeria has established a special unit to systematically use new AI models for tactical planning, weapons support, and propaganda.
New research published by the Cambridge Programme on AI Science & Policy (CASP) from Cambridge University has revealed concerning evidence that Boko Haram, a terrorist group in Nigeria, is fully utilizing Large Language Models, or Frontier AI, in its military operations. This contradicts the prior belief that armed groups are slow to adapt to new technologies. The research, led by Dr. Antonia Juelich, is based on 57 face-to-face interviews with 27 former Boko Haram members in northeastern Nigeria. The findings indicate that both main factions of the group have used readily accessible AI tools such as ChatGPT, Claude, Gemini, Grok, Meta AI, and DeepSeek for a variety of missions, ranging from tactical attack planning, gathering information for manufacturing explosives and drones, and propaganda, to conducting after-action reviews to refine their strategies. Notably, Boko Haram is not using AI in a scattered manner but has established a dedicated “AI unit” operating behind the front lines. This unit is responsible for prompting various AI systems, filtering the generated results into actionable advice, and forwarding it up the chain of command. There are also reports of foreign jihadist experts training senior commanders on how to use and bypass AI safety measures. Methods for circumventing the guardrails set by technology companies were another key finding. Former members revealed that the group uses a technique called “mixing and matching,” which involves combining answers from multiple AIs to compile complete information for dangerous objectives. For example, if one AI refused to provide an explosive formula but gave information about chemical components, the group would then use that information to ask another AI for mixing procedures. This trend is further corroborated by a parallel report from the Center for Strategic and International Studies (CSIS) in the United States, which indicates that terrorist groups are using multi-AI blending techniques to overcome protective measures.
This case represents some of the first concrete evidence confirming that terrorist groups can fully and systematically apply modern, accessible AI in real-world operations, posing a significant challenge to the safety measures of technology companies and security agencies worldwide.