Jul 13, 2026 19:05:00

It has been revealed that the Islamic extremist group Boko Haram was using AI to design explosives and repair weapons.

It has been revealed that

Boko Haram , an Islamist extremist group based in Nigeria, was asking AI how to make explosives and how to attack enemies. Industry insiders are concerned about the threat posed by terrorist plots that utilize AI.

“God has helped us, and so will AI”: How the Terrorist Group Boko Haram Uses Frontier AI — CASP
https://casp.ac/reports/ai-enabled-terrorism

How Terrorist Groups Are Using AI to Gain an Edge in Battle - The New York Times
https://www.nytimes.com/2026/07/10/us/politics/ai-terrorism-boko-haram-nigeria.html

Antonia Jülich, who leads an AI-related security program at Cambridge University, interviewed 27 former Boko Haram members in northeastern Nigeria to learn about their experiences using AI within the organization.

According to former members, Boko Haram uses cutting-edge AI such as ChatGPT, Claude, Gemini, Grok, Meta AI, and DeepSeek to carry out combat and daily operations, and even know-how on how to bypass the security measures set up by AI companies is circulating.

One example of its use was in modifying motorcycles. According to reports, members attempting to attack a military base surrounded by bunkers consulted the AI about how to jump over holes on motorcycles, referencing 'a scene from a movie where a motorcycle jumps over a bridge.' The mechanics modified the motorcycles according to the AI's instructions, practicing jumps that sometimes cost them their lives, until they were able to jump high enough to successfully carry out the attack, according to former members. In addition, they reportedly used the AI to learn how to design explosives and how to repair and modify weapons.

One former member said, 'You can type in a question or use voice input, and you'll get a detailed answer to questions like, 'How do you make a bomb?' and it will teach you how to make one. It's like a humanoid robot! We used it a lot.'

Companies that provide AI systems impose safety restrictions on their AI, and most are designed to reject harmful questions such as how to make a bomb. However, Boko Haram reportedly uses a combination of various AI systems to circumvent these restrictions. As can be seen from the aforementioned question about how to modify a motorcycle, which is not inherently harmful, there are also cases where they use AI by changing the way they ask questions.

Google, which had received a briefing from Jülich before the findings were released, stated, 'We have reviewed the content, but the answers were not specific or detailed enough to lead to misuse,' adding, 'Google has a strict policy against using AI for the purpose of causing harm in the real world.'

Anthropic, who also received a briefing from Mr. Jülich, said, 'Our products are designed to reject dangerous demands related to violence, attack plans, and explosive manufacturing. No company can combat these threats alone, so we work with external experts, researchers, and industry partners.'

A study conducted by Tech Against Terrorism, a UN-backed counter-terrorism non-profit organization, on more than 20 AI models showed that only 57% of thousands of terrorism-related questions were completely rejected. While approximately 80% of questions about explosives were rejected, only about one-third of questions about improvised chemical weapons were rejected.

Governments around the world are concerned that AI could be exploited to exploit system vulnerabilities, but there are also concerns that it could be used for terrorist activities. Jülich stated that 'the use of AI in terrorism is severely underestimated in both scale and nature,' and urged policymakers and AI developers to be vigilant.