IBM unveils defense-focused AI Model built to transform military Intelligence operations

IBM has introduced a breakthrough artificial intelligence solution specially designed for defense and national security uses. Announced on October 29, 2025, the IBM Defense Model is a strategic leap forward in how military organizations can benefit from AI technology for mission-critical decision-making while holding the highest level of security.

Developed through a strategic collaboration with Janes, a globally recognized authority in open-source defense intelligence, this purpose-built AI model combines IBM’s enterprise-grade watsonx.ai platform capabilities with specialized military domain expertise. Unlike traditional large language models that are trained for general consumer use, IBM Defense Model is trained specifically for defense operational tasks and can be used in air-gapped, classified and edge computing environments where connectivity to the internet is limited or disallowed.

Military command center utilizing AI technology for real-time threat analysis and mission planning 

Enterprise AI Governance Meets Military-Grade Security

What distinguishes this defense AI solution from competitors is its foundation on IBM’s Granite foundation models, which recently achieved a historic milestone as the world’s first open-source AI models to earn ISO 42001 certification for AI governance. The new international certification standard, launched in 2023, sets high standards for AI management systems in areas such as ethics, transparency, safety, and organizational responsibility.

The certification was completed by an accredited body, Schellman, through rigorous audits of the control policies and practices related to AI system governance, data handling practices, risk management and operational safeguards at IBM. For defense agencies working under strict compliance standards, this certification ensures external assurance that the AI model meets globally recognized standards for responsible AI development and deployment.

“Defense organizations need AI they can trust—solutions that deliver accurate insights without compromising security or ethics,” stated Vanessa Hunt, General Manager of Technology for IBM’s U.S. Federal Market. “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”.

Tackling the Hallucination Challenge in High-Stakes Environments

One of the most important technical innovations in the IBM Defense Model targets a major Achilles heel that has plagued AI systems: hallucinations-i.e., when AI models produce plausible-sounding but factually incorrect pieces of information. In military applications where decision-making can be a matter of life and death, the reliability of an AI-generated intelligence is a non-negotiable.

IBM approach focuses on teaching the model how to interpret real-time data from trusted sources—functioning like an intelligence analyst—rather than relying solely on memorized static datasets. The model integrates Retrieval-Augmented Generation (RAG) techniques, which research demonstrates can reduce hallucinations by 42-68% in AI systems. In some specialized applications, RAG paired with verified databases has achieved factual accuracy rates up to 89%.

The model is constantly being revised with intelligence updates via an integration with Janes dynamic defense intelligence database, ensuring that the model remains operationally relevant as the global threat landscape changes. This continuous learning architecture allows the system to stay relevant to the mission without giving up the accuracy that defense planners need.

Military personnel are monitoring AI-driven global operations and data in a command center. 

Air-Gapped Deployment: AI Without Internet Dependence

Modern defense operations increasingly occur in Denied, Disrupted, Intermittent, and Limited (D-DIL) environments where traditional cloud-based AI solutions are impractical or prohibited. The IBM Defense Model’s architecture supports deployment in completely air-gapped environments, secure systems physically isolated from unsecured networks and the internet.

This capability is especially important for classified military operations, for strategic command centers, and for forward-deployed units operating in hostile or contested territories. Recent demonstrations of air-gapped AI solutions during Exercise Mobility Guardian 2025 in the Indo-Pacific demonstrated the capabilities of portable “cloud-in-a-box” systems to provide secured, AI-powered collaboration at the tactical edge.

The IBM Defense Model is built on the zero-trust security principles that provide controlled user access across communication channels and allow for hybrid-cloud flexibility. Staff members will have the ability to use AI support for language translation, document enrichment, optical character recognition and intelligence synthesis without external internet connections that could open up vulnerabilities.​

Mission-Critical Use Cases Across Defense Operations

The IBM Defense Model is designed to support multiple high-value defense applications:

Intelligence Analysis and Reporting: Defense analysts can quickly process large amounts of operational data, extracting precise intelligence from military doctrine, threat assessments and real-time intelligence feeds. With its understanding of the domain-specific military language and operational understanding, the model is able to generate detailed, data-fueled reports for situation analysis.

Mission Planning and Strategy Development: Military planners can use the model for defense planning situations, wargaming simulations and in preparing strategic documents. By making the data collection and synthesis phases go faster, analysts can focus more cognitive resources on higher-value strategic thinking and decision making.

Document Enrichment and Knowledge Management: The system can supplement existing military documentation, provide cross-references to policies and procedures, and keep searchable databases of field manuals and operations instructions. This ability becomes particularly useful when field personnel need immediate access to repair manuals, tactical procedures or regulatory compliance documentation.

Scenario Modeling and Risk Assessment: AI-assisted scenario modeling allows defense teams to anticipate operational challenges, consider multiple courses of action and plan ways to mitigate risks before putting forces in the field.

The Technology Stack Powering Defense AI

The IBM Defense Model works as part of the wider watsonx AI platform ecosystem, which offers enterprise-level tools for building, training, validating, and deploying AI models throughout the full model lifecycle. The watsonx.ai studio gives access to AI developers in the form of prompt engineering tools, tuning capabilities, and data science toolsets powered by foundation models.

IBM Granite models use a decoder architecture that facilitates efficient language processing while requiring much less computational resources than larger competing models. The efficiency design means that Granite models can run on single GPUs and still deliver good performance on specialized business tasks, such as summarization, question-answering and classification.

The platform can be deployed across hybrid and multi-cloud environments, which helps defense organizations keep AI capabilities close to their data sources while overcoming data gravity challenges. For agencies that have personally identifiable information (PII) or classified data that need to be stored on-premises, watsonx.ai can be deployed in disconnected or air-gapped deployments while models are served on-premises, in the cloud, or operationally at tactical edge locations.

Industry Response and Future Implications

“Our collaboration with IBM brings together Janes trusted defense intelligence with IBM’s advanced AI capabilities,” Blake Bartlett, CEO of Janes, said. This model helps ensure defense organizations can get timely relevant insights in secured environments, helping them make informed decisions with confidence”.

The launch of the IBM Defense Model reflects broader trends in defense technology, where AI capabilities are rapidly becoming strategic assets for military organizations globally. As potential adversaries develop their own AI-powered capabilities, the ability to access accurate information quickly, safely, and securely, while maintaining confidence in the information’s veracity, represents a critical competitive advantage.

The model’s release also marks increasing enterprise adoption of industry- and use-case specific fit-for-purpose AI models, not just general-purpose large language models. This domain-specific approach allows organizations to achieve better performance on specialized tasks while having closer control over data security, model behavior, and compliance requirements.

Availability and Next Steps

The IBM Defense Model is now generally available to defense and national security agencies. Organizations interested in deploying the solution can access detailed information through IBM’s watsonx platform or engage with IBM Consulting services to accelerate AI integration into existing operational workflows.

As artificial intelligence keeps reshaping the defense technology landscape, solutions that harmonize cutting-edge capabilities with trusted governance frameworks will probably guide the next generation of military intelligence operations. With its specialized training, certified governance, and secure deployment architecture, the IBM Defense Model represents a significant milestone in the evolution of defense AI systems.

About Author

Netanel Siboni is a technology leader specializing in AI, cloud, and virtualization. As the founder of Voxfor, he has guided hundreds of projects in hosting, SaaS, and e-commerce with proven results. Connect with Netanel Siboni on LinkedIn to learn more or collaborate on future project.