Raw LLM Responses

DoD wants to play Calvinball. In July 2025 (Trump 2) Anthropic and DoD came to a deal with the two carveouts that Anthropic wanted - no domestic surveillance and no autonomous weapons with no human on loop. The first is a principled stand - domestic surveillance via AI is a run-around the 4th Amendment given the amount of publicly available information about individuals that can be purchased and gathered from data brokers. The second is a technological limitation - Claude is not ready to go to war and act as an autonomous weapon with no human oversight. Maybe it can get there, maybe it can't. In this thread, I'm specifically objecting to OpenAI publicly declaring solidarity with their competitor's position on Friday morning and then signed a deal to replace Anthropic on Friday night, hours after the administration made an example of Anthropic for holding that exact position.