Google is back in the defense business with a secret new Pentagon deal

Source: Android Authority

TL;DR

• The U.S. Department of Defense is reportedly tapping Gemini for classified projects with minimal restrictions from Google.
• The deal allows the Pentagon to use Gemini for “any lawful government purpose.”
• Inside Google, backlash is building again, with hundreds of employees warning about real‑world harm from AI misuse.

Pentagon’s New AI Agreement

The U.S. Department of Defense has decided to rely on Google Gemini for classified projects. A new agreement between the search giant and the Pentagon reportedly gives the latter full access to Google’s AI models, and Mountain View will not have any say in how the technology is used.

The Information reports that a source familiar with the deal says the contract lets the DOD use Gemini for “any lawful government purpose.” Google can recommend restrictions—such as not using the AI for autonomous weapons or domestic mass surveillance without human oversight—but the government is not obligated to follow these suggestions.

Context and Recent Developments

• Anthropic blacklist: Two months earlier, the Pentagon blacklisted Anthropic, officially due to supply‑chain risks. In practice, Anthropic would not allow its Claude AI to be used for certain military purposes, such as autonomous weapons and mass surveillance. The DOD and former President Trump indicated Anthropic could be reconsidered in the future, but for now the company remains excluded.

• Pentagon’s vendor strategy: Cameron Stanley, the Pentagon’s chief digital and AI officer, told CNBC that “overreliance on one vendor is never a good thing.” The DOD is now depending heavily on Google, OpenAI, and xAI for classified projects. Stanley says Gemini is already saving “thousands of man‑hours on a weekly basis” for U.S. military personnel.
  Source: CNBC

Employee Backlash at Google

Inside Google, more than 700 employees recently signed an open letter to CEO Sundar Pichai, urging the company to refuse classified projects. The letter warned, “Human lives are already being lost… from misuses of the technology we are playing a key role in building.”
Open letter (archived)

Historical Parallel: Project Maven

The situation echoes what happened in 2018 with Project Maven. Strong internal opposition then led Google to withdraw from a drone‑imaging contract and to create strict AI Principles.
More on Project Maven

Outlook

In 2026, with OpenAI and Elon Musk’s xAI already making similar agreements, Google likely feels pressure not to miss out on a major contract and the future of national‑security technology. The tension between commercial AI development and ethical concerns about military use continues to shape the industry.