Unilateral disarmament doesn't work though. If Anthropic is worried about this, just letting OpenAI win does seem genuinely worse.