While Anthropic sues the Pentagon and OpenAI faces employee backlash, Google is quietly becoming the Defense Department’s go-to AI provider. Last week, Google launched eight Gemini AI agents across the Pentagon’s unclassified networks, making its technology available to more than three million DoD employees.
The Deployment
The rollout went live on GenAI.mil, a government-run platform for AI tools. Google’s initial offering includes eight pre-built agents handling tasks like summarizing meeting notes, creating budgets, and checking proposed actions against national defense strategy guidelines. A no-code “Agent Designer” tool lets DoD employees create custom agents without programming knowledge.
For now, these agents operate only on unclassified networks. But the deployment positions Google as the default AI infrastructure for day-to-day Pentagon operations — a massive foothold that neither OpenAI nor Anthropic currently holds.
The Timing
Google’s expansion comes as its competitors are publicly fighting with the Pentagon:
Anthropic filed two federal lawsuits against the Trump administration on March 9 after being labeled a “supply chain risk” — the first time an American company has received that designation, historically reserved for foreign adversaries like Huawei. The Pentagon’s issue: Anthropic won’t allow Claude to be used for mass surveillance of Americans or fully autonomous weapons without human oversight.
OpenAI signed a Pentagon deal earlier this year, but lost its robotics lead Caitlin Kalinowski and several other employees who objected to military applications. The company now faces organized internal opposition to defense work.
Google, meanwhile, negotiated quietly and deployed without public controversy about terms or restrictions.
What Google Agreed To
Google has not publicly disclosed what guardrails, if any, apply to Pentagon use of Gemini agents. Unlike Anthropic, the company has not announced any blanket restrictions on surveillance or autonomous weapons applications.
This silence isn’t accidental. Google learned from the 2018 Project Maven backlash, when employee protests forced the company to abandon a Pentagon drone imagery analysis contract. This time, Google appears to be avoiding public ethical commitments that could trigger similar internal opposition.
But it’s not working entirely. Over 100 Google AI employees sent a letter to Chief Scientist Jeff Dean on March 1 requesting prohibitions on military use of Gemini for surveillance of Americans or autonomous weapons without human oversight — essentially the same restrictions Anthropic demanded and was blacklisted for.
The Broader Picture
More than 30 researchers from OpenAI and Google DeepMind, including DeepMind Chief Scientist Jeff Dean, filed a legal brief supporting Anthropic’s lawsuit. These aren’t random employees — they’re senior technical staff who understand what these AI systems can do.
The Pentagon has made its position clear: it wants AI “for all lawful purposes” and won’t accept restrictions from private companies. Defense Department CTO Emil Michael said Anthropic’s Claude models would “pollute” the defense supply chain because they have “a different policy preference.”
Who Wins
The immediate winner is Google. While competitors fight over ethics in public, Google deployed working AI agents to millions of Pentagon users. That’s three million people building habits around Gemini, three million potential advocates for expanding Google’s presence to classified networks.
The longer-term question: can Google maintain employee support while avoiding Anthropic’s restrictions? The letter to Jeff Dean suggests internal pressure is building. At some point, Google may face the same choice Anthropic did — draw ethical lines and risk Pentagon contracts, or accept any lawful use and risk employee backlash.
Anthropic bet that ethical limits were worth losing government business. Google is betting it can have both. The next few months will show which bet pays off.