Can we trust OpenAI and Anthropic to monitor their own technology? - Fortune