Google expands defence ties with Pentagon AI agreement

Alphabet’s Google has joined a growing list of technology firms partnering with the US Department of Defence to supply artificial intelligence models for classified operations, The Information reported on Tuesday.

The agreement allows the Pentagon to use Google’s AI systems for “any lawful government purpose”.

This places Google alongside firms such as OpenAI and Elon Musk’s xAI, which have also signed agreements to provide AI models for classified use.

AI to support sensitive defence operations

Classified networks are used to handle a wide range of sensitive government activities.

These include mission planning and weapons targeting.

The deal signals the Pentagon’s continued push to integrate advanced AI tools into critical defence workflows.

The Pentagon signed agreements worth up to $200 million each with major AI labs in 2025.

These included companies such as Anthropic, OpenAI, and Google.

The move reflects the department’s effort to retain flexibility in deploying AI systems without being restricted by limitations set by developers regarding the technology’s potential risks.

Flexibility and safety clauses in agreement

The report states that Google’s agreement includes provisions requiring the company to adjust its AI safety settings and filters at the government’s request.

This suggests that the Pentagon may seek modifications to existing safeguards depending on operational needs.

The contract also outlines certain limitations.

It states that “the parties agree that the AI System 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 agreement cited in The Information report, also makes clear that such provisions do not limit government authority.

Existing ties and contract amendment

A spokesperson for Google Public Sector told The Information that the new agreement is an amendment to an existing contract.

This indicates that Google already had an ongoing relationship with the US government for AI-related services.

Pentagon pushes AI firms for broader access

The Pentagon has been urging leading AI companies to make their tools available on classified networks.

This includes firms such as OpenAI and Anthropic.

The department has sought access to these systems without the standard restrictions typically applied to commercial users.

This approach highlights growing tensions between safety concerns raised by AI developers and the operational demands of defence agencies.

The latest agreement with Google underscores how major technology companies are increasingly becoming part of national security infrastructure, as governments look to harness AI capabilities for strategic and operational advantages.

Druckenmiller boosts Alphabet stake on AI strength

Meanwhile, Stanley Druckenmiller has increased his stake in Alphabet, citing its ability to control the full AI lifecycle.

He highlighted Google’s vertically integrated model, where it owns its stack from silicon to software.

Its in-house Tensor Processing Units (TPUs) provide a cost advantage in training Gemini models and improving search rankings.

This also helps avoid semiconductor supply volatility while boosting margins on AI inference.

The integration supports Google Cloud’s growth as AI firms shift workloads.

Druckenmiller described GOOGL as a “compounding machine” with a hard-to-replicate ecosystem.