Why Are Google DeepMind Employees Resigning Over Military AI Contracts?

Several technology workers have quit Google DeepMind, Google's artificial intelligence (AI) research division. They say they are deeply worried about Google's growing military contracts with the United States and Israel. They believe the company's AI technology could be used in ways that are unsafe or go against their values.

The resignations come after agreements made earlier this year that allow the Pentagon, the headquarters of the U.S. military, to use commercial AI systems for secret military work. Defense officials want to use AI to make the military more advanced. But the engineers who are leaving say these AI systems are not reliable enough and do not have enough human supervision to be used safely in dangerous military situations.

Ethical Concerns Lead Employees to Quit

Sofia Liguori, a research engineer based in London, left Google DeepMind this week after working there for three years. She said she decided to leave because Google was becoming more involved with military organizations.

"Realising my company is pushing more and more for AI systems in military operations makes me feel ashamed to work at Google," Liguori said. She spoke for herself, even though her employment contract includes rules that limit what she can share publicly. She said Google's new direction made her less confident that the company was following its own AI principles, which focus on protecting human rights and keeping people in control of AI systems.

Andreas Kirsch, another AI researcher who left DeepMind earlier this month, shared similar worries. He said AI systems made for ordinary business use are not suitable for use in combat.

"Google's governance and oversight are not sufficient or meaningful enough to deploy AI safely and responsibly in military and classified settings," Kirsch said. He also said Google seems to have little ability to see or stop how the U.S. Department of Defense uses its technology. He warned that today's AI models are not dependable enough to be used on battlefields.

The Military Wants to Put AI First

The disagreement at Google is happening as the U.S. government works to bring more advanced computer technology into military operations. U.S. Defense Secretary Pete Hegseth wants the military to start using AI quickly. He has said he wants the armed forces to become an "AI-first warfighting force," meaning AI would play a major role in how the military fights.

Under Secretary Emil Michael supported this approach earlier this year. He said that "speed defines victory in the AI era," meaning that being faster could be especially important for winning when AI is involved.

Google is not the only technology company working with the military. Other major companies, including OpenAI, Microsoft, Amazon, Oracle, Nvidia, SpaceX, and Reflection, have also signed contracts that allow their software to be used for moving supplies, planning military operations, and carrying out tasks in the field.

Much of this work is connected to systems such as Maven Smart Systems, a major platform developed by Palantir. It brings together and studies large amounts of information. It also uses AI technology from other companies to help military workers understand intelligence information and choose targets.

Can Humans Really Check AI Decisions During Combat?

Google's agreement allows its technology to be used for "any lawful government purpose," meaning any government activity that follows the law. However, the agreement also says AI models should not be used for weapons that act on their own or for large-scale government monitoring of people inside the country without proper human supervision and control.

Some defense experts and former technology leaders are worried that people may not be able to supervise AI properly during the pressure and confusion of combat.

Sarah Shoker, who previously led OpenAI's Geopolitics team, warned that AI can make military decisions happen much faster. "When you have hundreds of targets that are selected by AI and a short amount of time, it is almost impossible for a human operator to meaningfully review the AI's outputs for errors," Shoker said. In other words, when AI picks hundreds of possible targets very quickly, a person may not have enough time to check whether the system has made mistakes.

Retired Lieutenant General Jack Shanahan, the first director of Project Maven, raised a similar concern. He warned that the rush to use AI systems could mean skipping some of the careful testing and strict safety procedures normally required for military equipment.

Google Employees Have Protested Military Work Before

This is not the first time Google employees have been unhappy about the company's military projects. In 2018, employee protests led Google to stop its original work on Project Maven. The company then created its official AI Principles, which set out rules for how it should develop and use artificial intelligence.

More recently, Google's work with Amazon on Project Nimbus caused another controversy. Project Nimbus is a $1.2 billion contract to provide cloud computing and AI technology to the Israeli government. The project led to employee protests, and some workers faced disciplinary action from the company.

Some current Google employees spoke about the situation without giving their names. They said workers may be afraid to speak openly because their employment contracts have strict rules that discourage public criticism. Some also said that experienced researchers who disagree with military contracts are leaving, while new employees may simply take their places. Over time, this could change the company's workplace culture and what its employees consider ethically acceptable.

What Google Says

Google defended its military and government work. The company said it carefully checks these contracts to make sure they meet its rules and standards.

"We work with public sector organisations worldwide and follow a consistent process for reviewing and entering these contracts, including compliance with our acceptable use policies and AI principles," a Google spokesperson said. This means Google says it follows the same review process for government contracts around the world and checks that the agreements follow its rules about acceptable uses of technology and responsible AI.

Google also said it has special internal ways for employees to report worries about company behavior or ethical problems. The company said it has policies to protect workers from being punished for speaking up about these concerns.