Google strikes deal with the Pentagon, shreds prohibition on AI in weapons.
That one time that the Pentagon signed deals with 8 AI companies.
The deal includes Google, who kept employees in the dark about the deal, likely to avoid pushback. In 2018, Google signed a deal with the Pentagon, but ultimately backed out due to employee criticism.
At the time, Google added rules to their website regarding their AI policies, like not letting their tools be used for weapons. In Feb 2025, Google removed those rules from their website along with most of their internal communications, to prevent employees from organizing in opposition.
To make matters worse, Google’s deal was called the “most permissive,” and requires that Google remove safeguards when the Pentagon asks.
Anthropic was negotiating a deal with the Pentagon, but wanted provisions preventing use of their AI for domestic surveillance or autonomous drones, Pete Hegseth called them “woke” and labeled them a supply chain risk, leading Anthropic to sue the government. It’s likely that these 8 companies saw the writing on the wall if they didn’t comply with the Pentagon’s demands. But even Sam Altman, CEO of OpenAI, admitted that the deal was “sloppy” and opportunistic.
Source: @fubarpac