Google has officially shifted its stance on military partnerships, with top executives confirming last week that the company is “leaning more” into national security and defense contracts.
In a series of town hall meetings on March 20, 2026, Google and DeepMind leadership sent a “clear message” to employees: Google is aggressively rebuilding its relationship with the U.S. Department of Defense (DoD), marking a total reversal of the company’s 2018 decision to withdraw from military work.
The “GenAI.mil” Expansion
The most significant development is the massive rollout of Google’s Gemini for Government across the Pentagon’s new AI portal, GenAI.mil.
- Scale: The system is being deployed to approximately 3 million military and civilian personnel.
- Capability: Using a new “AI Agent Builder,” staff can create custom, no-code AI agents for unclassified tasks such as drafting contracts, summarizing intelligence reports, and project planning.
- The Goal: Undersecretary of War Emil Michael stated the intent is to create an “AI-driven culture change” that will eventually expand from unclassified tasks to Classified and Top Secret environments.
The Policy Shift: Principles Updated
To facilitate these deals, Google quietly updated its AI Principles in 2025, removing previous red lines that strictly prohibited the use of its technology for weapons or surveillance.
- The New Standard: VP of Global Affairs Tom Lue informed staff that the guiding principle for defense deals is now whether the “benefits substantially exceed the risks.”
- Target Areas: Google is specifically pursuing new contracts in cybersecurity, biosecurity, and autonomous logistics.
- Targeting Distinction: Leadership stressed that while they are leaning into defense, their tools are currently focused on administrative and “clerical” tasks rather than “identifying or striking targets.”
The “Anthropic” Vacuum
Google’s aggressive push coincides with a public breakdown between the Pentagon and rival Anthropic.
- The Dispute: Anthropic’s CEO, Dario Amodei, refused to remove safeguards against “mass domestic surveillance” and “fully autonomous weapons.”
- The Fallout: The Pentagon terminated Anthropic’s contracts, designated the company a “supply chain risk,” and is now facing a lawsuit from Anthropic.
- Google’s Opportunity: As Anthropic exited, Google Cloud CEO Thomas Kurian met privately with Pentagon officials on February 26 to pitch Google as a “reliable, expanded” alternative.
Employee Backlash 2.0
The move has reignited internal tensions within Google:
- “QuitGPT” Sentiment: Some employees have joined a broader tech-sector protest against military AI, while over 100 Google researchers signed a legal brief supporting Anthropic’s lawsuit against the government.
- Leadership Reassurance: DeepMind CEO Demis Hassabis stated he is “very comfortable” with the balance, arguing it is a responsibility to work with “democratically elected governments” to make the world safer.
