Frontier AI Evaluation Becomes Institutional
Anthropic's permanent third-party evaluator proposal, OpenAI's public agreement to pace frontier development and discussion of METR as a de facto standards body point to an external assurance layer between model release and public trust.