Google and OpenAI Accelerate Classified AI Negotiations as Pentagon Diversifies Beyond Anthropic Claude with Multi-Vendor Strategy

The Pentagon has accelerated negotiations with Google and OpenAI to deploy their AI models in classified military systems as the Defense Department pursues a multi-vendor strategy following the Anthropic standoff, according to sources familiar with the discussions.

Google's Gemini is reportedly "close" to reaching an agreement for classified deployment, while OpenAI's progress toward classified ChatGPT access has intensified in recent weeks despite being described as "not close" to finalizing terms due to complex contractual issues.

The renewed urgency follows the Pentagon's realization that relying on a single AI provider for classified operations creates unacceptable vulnerability. With Anthropic potentially facing "supply chain risk" designation over its refusal to remove safety guardrails, Defense officials are racing to establish alternative capabilities before any disruption to current operations.

A Defense official disputed earlier characterizations that Google was nearly finished while OpenAI lagged far behind, insisting that "discussions with both companies are ongoing and we expect both to eventually sign agreements" provided they accept the mandatory "all lawful purposes" requirement.

Both Google and OpenAI face the same demand that has created the Anthropic crisis: agreeing that their models can be deployed for any purpose the Pentagon considers lawful, including applications the companies might find ethically problematic such as autonomous weapons targeting or mass surveillance operations.

One source familiar with OpenAI's position said "it's unclear whether OpenAI would accept that condition, but we're talking," suggesting active internal debate over whether to follow xAI's accommodating approach or maintain usage restrictions similar to Anthropic's stance.

OpenAI softened its military-use policies in 2024, removing explicit bans on "military and warfare" applications in favor of narrower prohibitions on "weapons development." The shift signaled willingness to accommodate government contracts while maintaining some ethical boundaries, though the Pentagon's current demand would require abandoning those remaining guardrails.

Google has been less public about its AI ethics stance regarding military applications compared to Anthropic's vocal opposition. The company faced internal employee protests in 2018 over Project Maven, a Defense Department contract using AI for drone footage analysis, ultimately declining to renew that partnership.

However, Google's approach has evolved since then. The company maintained defense-related research partnerships and has competed for government cloud contracts, suggesting greater willingness to work with military customers than during the Maven controversy.

Sources indicate Google's hesitation centers less on principled opposition to military use and more on contractual complexity around liability, data handling, and performance guarantees in classified environments where standard commercial support models don't apply.

Both Gemini and ChatGPT already operate in the Pentagon's unclassified systems alongside xAI's Grok, providing Defense personnel with AI assistance for non-sensitive tasks. The classified deployment would extend access to systems handling intelligence analysis, weapons development planning, and real-time battlefield operations—dramatically higher-stakes applications requiring different technical architectures and security protocols.

Neither Google nor OpenAI has publicly commented on classified deployment timelines. Defense officials acknowledge that even after contractual agreement, technical integration requires months of security certification, system architecture modification, and operational validation before models can handle classified workloads.

The Pentagon's accelerated outreach to both companies signals that regardless of Anthropic's Friday deadline response, the Defense Department intends to establish redundant AI capabilities across multiple vendors to prevent future dependency on any single provider's cooperation.

For the AI industry, the Pentagon's multi-vendor push clarifies that classified military deployment represents a major revenue opportunity for companies willing to accept unrestricted usage terms, while creating competitive disadvantage for firms maintaining ethical guardrails that conflict with Defense requirements.