WASHINGTON: IBM has unveiled a new AI model built for the defense and national security sector — a purpose-built system designed to operate inside classified networks and deliver real-time intelligence without compromising security or ethics. Developed in partnership with Janes, the model represents a shift toward domain-specific AI that prioritizes trust, interpretability, and mission relevance.
IBM Steps Into the Defense Intelligence Arena
In a move that underscores the growing militarization of artificial intelligence, IBM on Tuesday announced the general availability of its IBM Defense Model, a large-language AI system developed specifically for defense and national security operations.
Unlike conventional AI models trained for broad civilian use, the Defense Model was designed from the ground up for mission-critical environments. It combines IBM’s enterprise-grade AI infrastructure with Janes’s defense intelligence data, long used by military analysts for open-source intelligence. The goal: to help defense organizations make faster, data-driven decisions within secure, isolated (“air-gapped”) networks.
“The IBM Defense Model provides a fit-for-purpose capability that accelerates mission planning and enhances operational readiness,” said Vanessa Hunt, General Manager for U.S. Federal Technology at IBM, adding that it reflects the company’s “commitment to responsible AI.”
Built for Security, Trained for Strategy
The Defense Model is built on IBM’s Granite foundation models, the world’s first open-source AI models to achieve ISO 42001 certification for AI governance. This certification, typically reserved for highly regulated industries, signals adherence to strict standards on transparency, oversight, and ethical deployment — a growing concern as AI becomes embedded in defense operations.
The model supports air-gapped and classified environments, meaning it can function independently of the internet, a key requirement for sensitive military use. According to IBM, this architecture ensures that no external data can infiltrate or exfiltrate secure systems — a safeguard that traditional, cloud-based AI models cannot guarantee.
Intelligence in Context, Not Just Data
Whereas most AI systems rely on memorized datasets, IBM’s approach focuses on teaching models how to interpret trusted, real-time information, much like a human intelligence analyst. This approach, informed by Janes’s defense-specific data and military doctrine, enables the model to understand operational language, contextual cues, and mission-specific decision frameworks.
The system will continuously update itself using Janes’s dynamic defense intelligence feeds, allowing agencies to access current situational insights for defense planning, analyst reporting, wargaming, and simulation. IBM officials argue that this “context-first” methodology helps reduce AI hallucinations — a persistent challenge in general-purpose language models — while maintaining accuracy under operational stress.
A Collaboration of Trust and Precision
For Janes, the partnership with IBM marks a convergence of trusted defense intelligence and governed AI innovation.
“Our collaboration with IBM brings together Janes’s trusted defense intelligence and IBM’s advanced AI capabilities,” said Blake Bartlett, CEO of Janes. “This model helps ensure defense organizations can access timely, relevant insights in secured environments, helping them make informed decisions with confidence.”
Industry observers note that the Defense Model also signals IBM’s broader strategy to align its AI products with national security imperatives — positioning it alongside defense-tech firms such as Palantir and Anduril, but with an emphasis on ethical, transparent, and standards-driven AI.
As global militaries race to adopt artificial intelligence, IBM’s entry into the field suggests that the next frontier of AI development may not be about scale or speed alone — but about trust, control, and governance in environments where every decision carries weight beyond computation.
