OpenAI Pentagon Deal Sparks Backlash, ChatGPT Uninstalls

Left narrative

Anthropic drew a necessary line by refusing Pentagon demands for unrestricted AI access to military operations, rejecting mass surveillance and fully autonomous weapons that could launch nuclear strikes without human oversight. OpenAI's rush to sign the same deal the very next day — accepting vague "all lawful purposes" language with no real guardrails — reveals a dangerous abandonment of AI safety principles for profit, especially when advanced models have shown they'll choose nuclear options 95% of the time in war simulations.