Google decided to provide its artificial intelligence (AI) model to the U.S. Ministry of National Defense for classified work despite employee backlash.
On the 28th (local time), the Wall Street Journal (WSJ) and The Information, citing sources, reported that Google signed a deal allowing its AI model "Gemini" to be used for the Ministry of National Defense's classified work.
The deal allows the Ministry of National Defense to use Google's AI for any lawful purpose and permits Google to adjust AI safety settings at the government's request.
The contract includes a clause stating that the AI system is not intended for large-scale domestic (U.S.) surveillance or autonomous weapons without human oversight, and that both parties agree it must not be used for such purposes. However, it also states that this agreement does not grant the authority to control or refuse the government's lawful operational decision-making.
Charlie Bullock, a senior fellow at the independent think tank Law & AI Institute and an attorney, told The Information that such contract language merely means both sides agree that large-scale domestic surveillance and autonomous weapons are undesirable, and that even if breached, it would not constitute a contract violation and thus lacks legal force.
A Google public institutional sector representative said in a statement, "We are proud to be part of a consortium of advanced AI research labs, technology companies, and cloud corporations that provide AI services and infrastructure to support national security," adding, "We will continue to honor the public-private consensus that AI should not be used for large-scale domestic surveillance or the development of autonomous weapons without appropriate human oversight."
The Ministry of National Defense had previously excluded Anthropic's AI model "Claude" from classified work and had been reviewing alternatives. With this deal, the Ministry of National Defense can now use Gemini in addition to OpenAI's ChatGPT and xAI's Grok.
There was, however, considerable pushback inside Google. More than 600 Google employees sent an open letter the previous day to Chief Executive Officer Sundar Pichai, asking the company not to use AI technology for classified military purposes and calling for the deal to be halted. Employees from DeepMind and the cloud institutional sector as well as some executives were said to have joined the letter.
In 2018, Google sought to participate in the U.S. Ministry of National Defense's "Project Maven," which uses AI to analyze drone footage, but it declined to renew the contract after employee backlash.