Unlike general-purpose AI models, the IBM Defense Model is built for defence-specific tasks and can be deployed in air-gapped, classified, and edge environments. It is based on IBM’s Granite foundation models and delivered through IBM watsonx.ai, aligning with the company’s strategy to offer smaller, domain-tuned open-source AI models that drive high-impact innovation.
“Defense organizations need AI they can trust – solutions that deliver accurate insights without compromising security or ethics,” said Vanessa Hunt, General Manager, Technology, U.S. Federal Market for IBM. “The IBM Defense Model provides a fit-for-purpose capability that accelerates mission planning and enhances operational readiness, while reinforcing IBM’s commitment to responsible AI.”
The model has been trained using military doctrine and data from Janes, enabling it to understand operational context and defence terminology. Rather than memorising static data, IBM’s method teaches the model to interpret trusted real-time information like an intelligence analyst, reducing hallucinations and improving relevance.
Key features include continuous updates from Janes’ dynamic defence intelligence data, support for secure deployments, and use in planning, reporting, document enrichment, wargaming, and simulation scenarios. IBM also highlights that its Granite models are the first open AI models to achieve ISO 42001 certification for AI governance.
“Our collaboration with IBM brings together Janes trusted defense intelligence and IBM’s advanced AI capabilities,” said Blake Bartlett, CEO for Janes. “This model helps to ensure defense organizations can access timely, relevant insights in secured environments, helping them make informed decisions with confidence.”
With this release, IBM aims to provide defence organisations with a secure and trusted AI tool that strengthens operational effectiveness and supports strategic decision-making in complex environments.
