In a groundbreaking study published in 2026 by the Cambridge Programme on AI Science & Policy, researchers reveal how factions of Boko Haram in Nigeria’s northeastern region have begun leveraging generative artificial intelligence to enhance their operational capabilities.
Beyond propaganda: How AI reshapes militant operations
The research, led by scholar Antonia Juelich, is based on 57 interviews conducted between 2025 and 2026 with 27 former members of the group. These included mid-level commanders and technical specialists, whose testimonies cover activities from 2023 to mid-2025. The study focuses on two main factions: the Islamic State’s West Africa Province (ISWAP) and Jama’atu Ahlis Sunna Lidda’awati wal-Jihad (JAS).
According to the report, these factions have adopted multiple publicly available AI tools—including ChatGPT, Claude, Gemini, Grok, Meta AI and DeepSeek—to streamline their activities. While AI was previously associated with propaganda production, the findings highlight far more strategic applications: planning attacks, resolving logistical challenges, optimizing troop movements and improving inter-unit coordination.
Former combatants revealed that AI systems were used to analyze failed operations by testing different attack scenarios, seeking explanations and identifying adjustments for future missions. Technical troubleshooting also benefited from these tools, with militants querying platforms to repair seized weapons, diagnose equipment malfunctions or understand military hardware operations.
The report notes that security measures built into AI platforms did not always prevent access to sensitive information. When denied responses, users reportedly reformulated queries, adjusted contexts or switched platforms to bypass restrictions—a practice that complicates monitoring efforts.
Structured AI units and Daesh-backed training
The study documents the creation of specialized AI units within these factions, composed of members skilled in engineering, weapons, explosives and strategic planning. Field operatives submit questions to these teams, which then query multiple AI systems before relaying vetted responses to combat units.
Access to computers, accounts and subscriptions is tightly controlled, limited to trusted members to prevent leaks or unauthorized use. This discipline reflects the influence of Daesh, whose instructors reportedly provided in-person training, remote support and even hardware—including laptops, projectors, encryption software and paid AI platform access—to ISWAP members.
The training covered AI fundamentals, query formulation and techniques to bypass platform restrictions. Insights gained were then disseminated across operational units, suggesting a deliberate effort to institutionalize AI use within the organization.
Critical gaps and evolving threats
Despite documenting these developments, the report acknowledges limitations. Findings rely on testimonies from former mid-level members—not current leadership—who may lack a comprehensive view of the group’s strategic decisions. Additionally, the study does not verify the accuracy of AI-generated responses in real-world conditions, as these models can produce incorrect, incomplete or impractical advice.
Still, the research underscores a stark reality: armed groups with limited resources can now access the same AI tools as governments, corporations and individuals. This accessibility lowers technical barriers, accelerating knowledge-sharing in translation, information analysis, maintenance and planning. While AI cannot replace human expertise, it significantly speeds up learning and operational adaptability.
For tech developers and security agencies, the challenge lies in detecting malicious use without stifling legitimate applications. The proliferation of platforms, accounts and circumvention methods complicates oversight, prompting researchers to call for stronger collaboration among developers, governments and security services.
The study concludes that Boko Haram factions already exhibit high confidence in AI’s potential, likely driving future investments in subscriptions, hardware and specialized training. The question is no longer whether these groups will use AI, but how effectively security mechanisms can adapt to this evolving threat.
