OpenAI and Anthropic Agree on the Safety Diagnosis — and Reach Opposite Conclusions About Going Public - FourWeekMBA