“Together, the War Department and these strategic partners share the conviction that American leadership in AI is indispensable to national security,” a Pentagon press release said, announcing a portfolio of cleared developers intended to avoid dependence on a single vendor.
Which companies are cleared
Pentagon officials said Friday that seven AI developers have deals to provide tools inside classified Defense Department networks. The named companies are Amazon Web Services, Google, Microsoft, NVIDIA, OpenAI, Reflection, and SpaceX. The statement emphasized the breadth of the group—“which does not include Anthropic”—as a deliberate hedge against “vendor lock.”
Impact Level 6 and Impact Level 7 environments
The clearance covers tools for Impact Level 6 and Impact Level 7 network environments. Pentagon officials described the move as part of a bid to streamline data synthesis, improve warfighter decision-making, and increase situational understanding and awareness. The announcement frames the access as operational: the cleared products will be available to support missions that operate at those classified impact levels.
GenAI.mil and Google’s Gemini 3.1 Pro
The new AI capabilities will be accessible via GenAI.mil, the Pentagon’s central AI platform. The source notes that in late April Google rolled out its Gemini 3.1 Pro model on GenAI.mil, demonstrating how at least one cleared provider has already integrated a specific model into the central platform intended for departmental use.
An apparent response to the Anthropic dispute
The announcement follows public tensions between the Pentagon and Anthropic. According to the source, those tensions “exploded in late February” after Anthropic refused to allow its products to be used for autonomous weapons and surveillance of Americans. The Pentagon subsequently designated Anthropic a supply-chain risk, and the president ordered federal agencies to begin offloading use of its products, though a judge has issued an injunction on those actions.
What this means for the War Department, model developers, and warfighters
- The War Department and the Department will have multiple cleared vendors to supply AI capabilities on classified networks, a structure the press release describes as supporting “the full and effective use” of models in Department missions. The release also invoked a policy direction “mandated by President [Donald] Trump and Secretary [Pete] Hegseth,” tying the rollout to senior-level directives.
- Model developers named in the clearance gain a formal pathway to serve Impact Level 6 and 7 environments on GenAI.mil; Google’s Gemini 3.1 Pro is an early example of that integration. For those developers, the clearance is an operational opening into a centralized, department-wide platform.
- Warfighters and mission planners stand to receive tools intended to streamline data synthesis and increase situational awareness. The Pentagon framed the capability expansion as strengthening decision-making at the unit and command level through access to multiple AI providers.
The Pentagon’s announcement packages a strategic argument—ensure access, avoid vendor lock, and move quickly to field capabilities inside classified enclaves—into a single, actionable step: clear a group of seven providers and deliver their models through GenAI.mil. It also leaves in plain view a recent fault line: Anthropic’s public refusal on certain use cases and the subsequent supply-chain designation, followed by judicial intervention, underscore that vendor access is as much political and legal as it is technical.
The immediate next fact on the record is that those seven providers are cleared for Impact Levels 6 and 7 and that GenAI.mil will host their tools; beyond that, the administration’s broader plan—how integration will be governed across the cleared vendors, how operational security will be enforced on classified networks, and how the department will manage disputes like the Anthropic case—remains to be articulated in further department releases or actions.
