IL7: The Most Stringent Cloud Security Classification

The Department of Defense has expanded its classified artificial intelligence initiatives by partnering with eight companies to deploy generative AI capabilities within its most secure cloud environments. This expansion specifically targets the use of Impact Level 7 (IL7) security classifications, the most stringent standard for cloud computing and hosting services.

While the initiative includes several of the largest names in cloud infrastructure and AI development, Anthropic was not among the companies selected for this specific expansion of classified work, according to reporting from DefenseScoop.

The IL7 Security Standard

The move to IL7 environments is a critical technical requirement for the Department of Defense. Unlike lower impact levels used for unclassified or secret data, IL7 is designed to host Top Secret and Sensitive Compartmented Information (SCI).

These environments ensure that generative AI models can process the government’s most sensitive intelligence and operational data without the risk of leakage to lower-security networks or the public internet.

By utilizing IL7 hosting, the Department of Defense can implement large language models (LLMs) and other AI tools in air-gapped or highly restricted cloud shells that meet the rigorous requirements of the intelligence community.

CDAO and GenAI.mil Integration

The effort is being driven by the Chief Digital and Artificial Intelligence Office (CDAO), which manages the integration of AI across the defense enterprise. The expansion is part of a broader strategy to operationalize generative AI through the GenAI.mil initiative.

GenAI.mil serves as the central framework for the Department of Defense to experiment with and deploy AI tools. The CDAO focuses on ensuring these tools are not only powerful but also compliant with strict data governance and security protocols.

Leadership within the CDAO, including officials such as Lauren Kahn and Emil Michael, have been central to coordinating how these commercial AI capabilities are transitioned into classified military applications.

Industry Participation and Exclusions

The selection of eight companies reflects a mix of cloud service providers and AI developers. The list includes firms capable of providing the massive compute power required for AI, such as NVIDIA, as well as those providing the cloud fabric, including Amazon Web Services (AWS), Microsoft, Google, and Oracle.

The exclusion of Anthropic is notable given the company’s prominence in the generative AI field and its focus on AI safety. However, the Department of Defense’s selection process for IL7 work typically depends on a company’s ability to meet specific facility clearances and the technical capacity to operate within air-gapped environments.

Other participants in the broader AI ecosystem, such as OpenAI and SpaceX, remain relevant to the Department of Defense’s overall technological roadmap, though the IL7 expansion focuses on a specific subset of partners capable of meeting the highest security thresholds.

Strategic Implications for Defense AI

The shift toward IL7 AI deployment indicates a move from general experimentation toward the actual operational use of generative AI in high-stakes intelligence environments. By integrating these tools into classified systems, the Department of Defense aims to accelerate the analysis of vast amounts of encrypted data.

This approach allows the military to leverage commercial innovation in AI while maintaining the isolation required for national security. The use of eight different partners suggests a strategy of avoiding vendor lock-in and ensuring redundancy across the defense cloud architecture.