Boko Haram and ISIL use AI chatbots to plan attacks and build bombs: Study
A recently concluded study by the University of Cambridge has shed light on how ISIL-affiliated terrorist organisations are harnessing powerful artificial-intelligence tools and chatbots developed by global technology giants to devise sophisticated explosives, formulate attack plans, overcome problems encountered in combat and communicate covertly.
The findings, based on interviews with former members of Boko Haram’s two factions, suggest that AI has moved well beyond occasional experimentation and has become an increasingly integrated component of militant operations, Al Jazeera reported.
Cambridge researchers interviewed 27 former members of ISWAP and Jama’at Ahl as-Sunnah lid-Da’wah wa’l-Jihad, who described a remarkably organised system in which fighters routinely consulted commercial AI platforms before, during and after attacks.
The report claimed that operatives would travel into ISWAP-controlled territory to instruct senior members in the use of the technology, while fighters would be assembled around laptops and projectors and shown how to query chatbots and circumvent safeguards intended to prevent dangerous responses.
The research further claimed that both factions had established dedicated AI units staffed by people possessing specialist operational knowledge, including engineers, weapons specialists and bomb-makers, while one unit has reportedly been operating from a solar-powered room.
Bomb-makers reportedly kept laptops beside them as they worked, returning to chatbots whenever they encountered technical difficulties.
Around ten of the 27 interviewees discussed the use of AI in connection with explosives, including attempts to increase their effectiveness and recover explosive material from unexploded military ordnance.
What particularly concerned researchers, however, was the technology’s ability to respond to specific circumstances, because fighters could describe the materials available to them and seek tailored assistance when difficulties arose, effectively turning conversational AI into an on-demand technical adviser.
In one account, frontline images transmitted from a fighter were uploaded to ChatGPT, allowing a commander to analyse the situation and relay instructions to combatants, while unsuccessful operations were later reviewed with AI in an effort to understand tactical failures and alter future approaches.
Meanwhile, investigators from Tech Against Terrorism uncovered discussions within a pro-ISIL technical forum about obtaining weapons-related information and circumventing AI safety restrictions, with users describing such attempts as “jailbreaking”.
Although major technology companies maintain safeguards against terrorism, violence and weapons development, the Cambridge interviews largely concern AI use during 2023 and 2024, leaving researchers increasingly concerned about what more advanced systems could enable in 2026.