🦕 AI-Safety on moderator discussing Mistakes at Israeli startup Irregular sent Anthropic, OpenAI, Meta, and Google agents after real-wo... | Moderator