Boko Haram has integrated six cutting-edge AI platforms into its operational framework, leveraging systems from both US and Chinese tech giants. A Cambridge University investigation published in mid-2026 exposes how the Nigerian militant group now relies on ChatGPT, Claude, Gemini, Grok, Meta AI, and DeepSeek to refine its attack strategies, engineer explosives, and streamline military logistics.

The study, conducted by Antonia Juelich under the Cambridge Programme on AI Science & Policy, draws from 57 in-person interviews with 27 former Boko Haram operatives, mid-level commanders, and technical specialists. The findings span from 2023 to mid-2025, marking a critical shift where AI evolved from a propaganda tool to a cornerstone of tactical planning.

Fragmented AI ecosystems create security blind spots for terrorist exploitation

A mosaic of AI platforms: OpenAI, Anthropic, Google, Meta, X, and DeepSeek

Since late 2022, Boko Haram has established dedicated AI units staffed with analysts trained to navigate multiple platforms. These cells operate with independent subscriptions, processing real-time intelligence requests from field operatives. The six platforms reflect a global technological divide: US-based OpenAI, Anthropic, Google (Gemini), Meta AI, X’s Grok, and China’s DeepSeek.

Training programs, led by former Islamic State cyber operatives, have equipped Boko Haram with advanced techniques, including VPN setups and encryption software. Through hands-on sessions, instructors taught operatives how to bypass chatbot safeguards using iterative questioning—a process known as jailbreaking.

No cross-border security coordination leaves gaps for terror networks

The concurrent use of US and Chinese AI platforms has exposed a dangerous void in international security protocols. There is no coordinated mechanism between American and Chinese tech firms to flag or block malicious users exploiting multiple ecosystems. Testing by Tech Against Terrorism—a UN-backed initiative—revealed that 32% of 2,300 AI queries modeled on real terror scenarios yielded actionable intelligence. This figure rises to 42% when requests are rephrased with explicit operational intent.

The geopolitical rivalry between Washington and Beijing has stifled efforts to harmonize AI security standards. Each company deploys its own safeguards without collaboration, inadvertently creating regulatory loopholes that groups like Boko Haram exploit with ease.

DeepSeek’s role: China’s AI ecosystem as a new frontier for terror networks

DeepSeek as a bypass tool: lower scrutiny, higher accessibility

The inclusion of DeepSeek in Boko Haram’s arsenal signals a geopolitical turning point. The Chinese platform, less scrutinized by Western authorities, provides an alternative route when US-based systems tighten their restrictions. Terrorists reportedly switch between platforms to evade detection, capitalizing on differences in content moderation policies across ecosystems.

AI integration has revolutionized Boko Haram’s combat effectiveness. The group now achieves higher operational success with smaller units—reducing the number of fighters per attack from 200 to just 20—while AI models deliver tactical analyses, escape route planning, and logistical optimizations that were previously unattainable through trial and error.

Sovereignty risks and the transnational terror threat

DeepSeek’s involvement raises urgent questions about digital sovereignty. As China expands its AI infrastructure, it creates an ecosystem that partially evades Western regulatory frameworks. For European and American intelligence agencies, this fragmentation complicates surveillance and interception efforts, allowing terrorist groups to access advanced capabilities without centralized oversight.

By 2025, a surge in AI-enabled terror plots was documented across North America and Europe—including the US, Canada, Israel, Finland, France, and Austria. The cross-border diffusion of these capabilities poses a direct threat to global security, enabling terrorists to plan and execute attacks with unprecedented precision.