That companies like OpenAI, Anthropic, Google, and others might simply unfreeze their model weights and reveal their system prompts—all in the name of transparency regarding the AI itself.