Google Permits U.S. Military Use of AI Models Amid Employee Opposition

Google has agreed to terms allowing its Gemini AI models to be used in U.S. The agreement follows similar deals by OpenAI, xAI, Nvidia, Microsoft, and Amazon. Anthropic refused such terms, leading the Pentagon to order a halt to its products' use within six months and label it a supply chain risk, which Anthropic is challenging in court.

In response to the deal, nearly 600 Google employees signed an open letter opposing it. The company issued a memo stating it proudly works with the U.S. military and intends to continue. Several employees have publicly criticized the agreement, including Alex Turner, a research scientist at Google DeepMind, who posted on X that Google cannot veto usage, commits to modifying safety filters at government request, and includes aspirational language without legal restrictions.

The researcher said many staff were unaware of the deal due to lack of clear communication from Google, and an internal memo on responsible AI did not explicitly address the agreement. The open letter protesting military use has now gathered around a thousand signatures, according to the researcher.

Concerns focus on AI's potential in autonomous weapons, where it could identify and select targets without human oversight, and in mass surveillance, aggregating data into comprehensive profiles, which legal experts say is currently lawful under U.S. laws including the 1978 Foreign Intelligence Surveillance Act, the 2015 USA Freedom Act, and the Fourth Amendment.

The deal raises questions about AI companies' ability to limit government use of their technology, particularly for weapons and surveillance, and the extent of employee influence over such decisions.