Google has quietly reached an agreement with the Pentagon to make its artificial intelligence models available for classified government work, the latest sign of how deeply Silicon Valley is becoming intertwined with U.S. national security.
According to The Information, the deal allows the Defense Department, recently renamed the Department of War by President Donald Trump, to use Google’s AI systems for “any lawful government purpose.” This puts Google alongside OpenAI and Elon Musk’s xAI, both of which have already secured similar arrangements.
The contracts, part of a broader 2025 push, are reportedly worth up to $200 million apiece with leading AI labs including Anthropic, OpenAI, and now Google.
Register for Tekedia Mini-MBA edition 20 (June 8 – Sept 5, 2026).
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
Register for Tekedia AI Lab.
The Pentagon has been actively pressing top AI companies to provide their most advanced models on classified networks with fewer restrictions than they typically impose on commercial customers. These systems are used for highly sensitive tasks, ranging from mission planning to weapons targeting.
Under the agreement, Google will be required to assist the government in adjusting the company’s AI safety settings and content filters when requested. The contract includes explicit language stating that the AI “is not intended for, and should not be used for, domestic mass surveillance or autonomous weapons (including target selection) without appropriate human oversight and control.”
However, the deal also makes clear that Google has no right to control or veto lawful operational decisions once the technology is in the Pentagon’s hands. That carve-out exposes the limits of private-sector influence when working with the military, which has fueled concern over growing deals between the Pentagon and AI companies for the use of advanced models.
Google defended the partnership in carefully worded terms. A company spokesperson told Reuters that Google supports government agencies on both classified and unclassified projects and remains committed to the industry consensus that AI should not be used for domestic mass surveillance or autonomous weaponry without meaningful human oversight.
“We believe that providing API access to our commercial models, including on Google infrastructure, with industry-standard practices and terms, represents a responsible approach to supporting national security,” the spokesperson said.
The Pentagon has consistently said it has no interest in using AI for mass surveillance of Americans or for fully autonomous lethal weapons, but it wants maximum flexibility for any lawful application.
The agreement marks a notable evolution for Google, which has historically been cautious about deep military involvement, most famously pulling back from Project Maven several years ago after internal employee protests. Its willingness to now sign on with fewer restrictions than some rivals signals a pragmatic shift as the race to deploy advanced AI in defense intensifies.
The move follows an ongoing fallout between Anthropic and the Pentagon. Earlier this year, Anthropic clashed with the Pentagon after refusing to remove guardrails designed to prevent its Claude models from being used in autonomous weapons or domestic surveillance. The standoff led the Defense Department to designate Anthropic a “supply-chain risk,” a serious blow for the startup.
By contrast, Google, OpenAI, and xAI appear more willing to accommodate the Pentagon’s desire for broader access in exchange for access to major government contracts and strategic relevance.
However, this latest development indicates that artificial intelligence has moved from a commercial curiosity to a core asset in great-power competition. As the U.S. seeks to maintain its technological edge, particularly against China, the Pentagon is determined to harness the most powerful commercial AI models rather than relying solely on slower, in-house development.
While the deal offers significant revenue potential and strengthens Google’s position in the rapidly expanding government AI market, it also carries risks, from potential employee backlash and reputational concerns to future ethical and regulatory scrutiny over how the technology is ultimately deployed.
As more AI companies embed themselves in the national security apparatus, the line between Silicon Valley innovation and military application continues to blur. The concern, thus, remains whether these carefully negotiated guardrails will hold in practice once the models are integrated into real-world classified operations.