Silicon Valley giant Google is edging closer to a major comeback with the US military — this time anchored in artificial intelligence (AI).

The company is in active talks with the Department of Defense to deploy its Gemini AI in classified environments, according to reports citing insiders.

If finalized, the arrangement could see the Pentagon using the tech for “all lawful purposes.”

Still, Google is pushing for clear safeguards, including restrictions on mass surveillance of Americans and limits on fully autonomous weapons without “appropriate” human oversight and control.

The move also signals more than just a software intiative. Sources said Google is quietly deepening its infrastructure play, beefing up its secure cloud capacity and adding high-end computing hardware needed to train and run AI systems at scale.

Those capabilities are considered critical if Gemini is going to operate inside top government systems.

Dramatic Pivot

The talks mark a sharp turn for Google, which walked away from a Pentagon drone AI project in 2018 following internal protests and concerns over its use.

Fast forward to now, the latest push came amid a messy fallout between the Pentagon and Anthropic, whose own discussions unraveled over concerns tied to domestic spying and rogue robotic technologies.

The fallout even led to the startup being flagged as a “supply chain risk.”

Without addressing the reported Google talks, a Pentagon official said the department will “continue to deploy frontier AI capabilities through strong industry partnerships.”