ISIL Fighters Use AI Chatbots to Build Bombs

Overview

A new study by the University of Cambridge reveals that armed groups affiliated with the Islamic State (ISIL) in Nigeria are systematically deploying artificial intelligence chatbots developed by major technology companies for military activities. Researchers found, through in-depth interviews with 27 former Boko Haram members, that the group not only uses mainstream tools such as ChatGPT, Claude, and Gemini, but also extensively leverages platforms like Grok to plan attacks, manufacture bombs, and solve combat challenges.

Meanwhile, exclusive material obtained by Al Jazeera shows that a broader base of ISIL supporters is sharing guides on technical forums to circumvent AI safety restrictions. This trend has raised deep concerns within the international community about the misuse of new technologies by extremist groups.

Training System and Organizational Structure

Within a rebel stronghold deep in northeastern Nigeria, former combatants described a highly organized and sophisticated technological application system.

• Centralized training: Operators travel to areas controlled by ISWAP, where external experts demonstrate to senior members how to use AI. Training is conducted via projector on large screens, with encrypted tools and laptops provided on site.
• Dedicated teams: Both ISWAP and JAS factions have established specialized AI units. These units are composed of personnel with expertise in engineering, weapons, and explosive device manufacturing; some teams number up to 23 members and operate in rooms powered by solar energy.
• Round-the-clock support: One former commander described AI as a “humanoid robot” available to answer questions at any time. Combat engineers often keep laptops nearby while assembling explosive devices, querying them immediately when problems arise.

“The answers we get cover everything,” said a former ISWAP commander. “God helps us, and so does AI.”

Combat Applications and Security Risks

The core issue identified in this study is that AI can provide customized solutions for specific situations, not just general information.

1. Improving explosives: Fighters inform the AI about the limited materials at hand and request optimization of explosive effects. The AI then provides 24-hour guidance akin to that of a technical expert, adjusting recommendations based on specific materials and targets.
2. Salvaging munitions: Regarding unexploded ordnance, ISWAP consults AI on how to safely recover military-grade explosives. After a previous accident caused by improper handling resulted in 40 deaths, such technical guidance is seen as a key means of reducing risk.
3. Drone modifications: The study also indicates that AI is used to modify commercial drones into attack platforms capable of carrying and deploying explosives.

Although some posts have been deleted, University of Cambridge researcher Antonina Juelich warns that this “query-feedback” model has given non-professional combatants unprecedented technical support capabilities, highlighting the fragility of current AI safety guardrails against malicious exploitation.