Report Claims Boko Haram Used AI Chatbots to Improve Bomb-Making, Attack Planning

Former members of the terrorist group Boko Haram have claimed that the organisation used artificial intelligence (AI) chatbots to improve bomb-making techniques, refine attack planning and solve operational challenges, according to a new research report that has drawn international attention.

The findings, featured in The New York Times and based on research by Antonia Juelich of the University of Cambridge’s Programme on AI Science and Policy, are drawn from interviews conducted with former Boko Haram members in northeastern Nigeria.

According to the report, former insurgents said the group relied on several publicly available AI platforms, including ChatGPT, Claude, Gemini, Grok, Meta AI and DeepSeek, to seek information that assisted combat operations, weapons troubleshooting and logistical planning.

Claims of AI-assisted operational planning

One former Boko Haram commander reportedly told researchers that the group turned to AI after an attack on a military base was frustrated by a defensive trench.

According to the account, fighters asked AI tools how motorcycles could be modified to clear such obstacles. The responses allegedly helped mechanics adjust the motorcycles for greater speed and acceleration before subsequent operations.

The former commander claimed the group repeatedly practised the manoeuvre before using it during an attack.

Bomb-making allegations

The report also quoted former members of the Islamic State West Africa Province (ISWAP), a Boko Haram splinter faction, as claiming they used AI chatbots to obtain information related to explosive devices after disguising their requests to avoid safety restrictions.

Researchers noted that these accounts are based on interviews with former fighters and should not be interpreted as independent verification of every operational claim.

The Cambridge study concludes that the interviewed former members described AI as being used across several aspects of terrorist activity, including operational planning, technical troubleshooting and internal training.

AI companies respond

Responding to the findings, OpenAI said the reported activities violated its usage policies, which prohibit using its services to facilitate violence or terrorist activities.

Google and Anthropic also stated that their AI models are designed with safeguards intended to refuse requests that could facilitate dangerous or harmful activities.

However, the former insurgents interviewed for the study claimed they were sometimes able to bypass those safeguards by disguising malicious requests as legitimate academic, engineering or fictional projects.

Growing security concerns

The researchers warned that the findings illustrate how extremist groups may seek to exploit widely available AI technologies for operational advantage, even as technology companies continue to strengthen safeguards against misuse.

The report calls for closer collaboration among governments, AI developers and security agencies to reduce the risk of advanced AI tools being exploited by terrorist organisations while preserving their legitimate civilian applications.