Dario’s narrative has become almost theological. First, AI may be conscious. Then AI becomes autonomous. Then AI becomes an existential threat. Then China must be denied advanced models and compute. Then humanity is told it may be approaching catastrophe. And finally: Here is the doctrine that will save you. The remarkable part is not that Anthropic talks about AI risk. It is that the same institution increasingly casts itself as the authority that gets to define the intelligence, define the danger, define the geopolitical enemy, and prescribe the salvation. At some point, this stops sounding like safety research. It starts sounding like a Silicon Valley church whose product, apocalypse and messiah all come from the same company.