Google DeepMind Staff Resign Over Military AI Contracts
Sofia Liguori, a London-based research engineer at Google DeepMind, has resigned after three years with the company. Her departure is part of a growing concern among employees about the company’s involvement in military artificial intelligence (AI). According to a BBC investigation, several staff members have either quit or expressed reservations about the role their work plays in developing AI for military use.
The main issue at hand revolves around whether Google can effectively monitor and control how its AI systems are used once deployed in classified environments. Employees like Liguori worry that human oversight may become meaningless in high-stakes decisions, where speed is of the essence. This concern stems from the potential risks associated with AI’s rapid processing capabilities, which could make it harder for operators to review outputs and identify errors.
Google has maintained that it follows a consistent process for reviewing public sector contracts, including compliance with its acceptable use policies and AI principles. The company also claims to have dedicated channels in place for employees to raise questions without fear of retaliation. However, critics argue that these measures may not be sufficient to address the complexities involved in military AI development.
The Pentagon’s expansion into AI has sparked a heated debate about accountability and human control in high-stakes settings. In May 2026, the Department of Defence announced agreements with eight AI companies, including Google, to deploy AI on its classified networks for lawful operational use. The contract reportedly allows the Pentagon to utilize Google’s AI for any lawful government purpose, but does not grant the company a right to control or veto decisions.
The employees’ concerns centre around how Google’s commitments to implementing human oversight and aligning with international law and human rights principles apply when supplying AI systems for classified military use. This raises questions about the reliability of AI in warfare without more testing, stronger guardrails, and meaningful human involvement.
Discussion
No messages yet. Be the first to write.