Google DeepMind staff resign over military AI contracts

Key points

• Sofia Liguori and Andreas Kirsch resigned from Google DeepMind, citing shame and insufficient governance regarding AI deployment in military settings.
• Google signed contracts with the US Pentagon earlier in 2026, allowing its AI models to be used for classified military work, including target selection with human oversight.
• The US Defense Department, led by Secretary Pete Hegseth, prioritizes becoming an 'AI-first warfighting force,' accelerating the integration of AI into platforms like Maven Smart Systems.
• Critics, including former OpenAI head Sarah Shoker, warn that AI speeds up decision-making beyond human capacity to review errors in high-stress combat environments.
• Google employees report fear of retaliation for speaking out, as employment contracts prohibit public criticism of the company, though some are leaving despite these restrictions.

Background

This development follows earlier internal tensions at Google regarding AI ethics, such as the 2024 controversy over Project Nimbus, a $1.2bn contract with Amazon providing AI infrastructure to Israel. Recent internal testing of the 'Carbon' model for coding also highlights the company's rapid pace of AI development, which some staff argue outpaces safety protocols.

Why it matters

The resignations signal deep internal conflict over the ethical boundaries of AI in warfare. As major tech firms integrate AI into defense systems, concerns about inadequate human oversight and the speed of automated decision-making in combat could have significant implications for international law and civilian safety.

What to watch

Google may face continued internal attrition as more employees leave or remain silent due to fear of retaliation. The Pentagon is expected to continue expanding AI use in military operations, potentially facing increased scrutiny from experts and the public regarding the safety and reliability of these systems.