“Ben Thompson of Stratechery theorizes that Anthropic's controversial decisions are driven by a sincere, internal belief that it is uniquely focused on and capable of managing AI safety.”