DeepMind Scientist Resigns Over Pentagon AI Deal

Corporate AI ethics pledges are meaningless public relations devices that dissolve instantly when confronted with government pressure and financial incentives.

• –Stated safety principles were quietly removed from Google's guidelines in 2025 to enable military contracting, demonstrating that internal governance lacks independent enforcement power.
• –The belief that safety can be secured by having a personal "seat at the table" failed to prevent the signing of a deal that lacks any binding ethical restrictions.
• –Senior AI leaders who pledged to never support autonomous weapons chose to remain at Google after it signed the contract, showing that their individual leverage was not utilized when it mattered most.
• –The Pentagon successfully used coercive pressure, such as threatening to designate Anthropic as a supply chain risk, to force AI providers to accept unrestricted use terms.