OpenAI and Anthropic share findings from a joint safety evaluation | Not Hacker News!