4 articles tagged with Pentagon
Newly released court emails between Dario Amodei and the Pentagon reveal the real fight — whether an AI lab can contractually refuse the state. The precedent will bind every vendor.
On April 28, 2026, Google signed a classified Pentagon AI deal on the same "all lawful uses" terms Anthropic refused for fourteen months — four days after committing $40B to Anthropic. The cascade isolates Anthropic as the industry's lone holdout on autonomous weapons and domestic surveillance carve-outs, and reveals that AI ethics positioning has become a priced asset class.
The complete story of how Anthropic refused to remove AI safety guardrails for autonomous weapons and mass surveillance, got designated a supply-chain risk to national security, and watched OpenAI sign a Pentagon deal with identical protections hours later. Timeline, analysis, and what it means.
The Pentagon has given Anthropic 48 hours to strip safety guardrails from Claude or face blacklisting, contract termination, and wartime production law. This is the most consequential confrontation between AI safety principles and state power in history.