While Altman recently suggested society may need to accept "some bad things happening" as AI evolves, Huang says zero tolerance should be the rule. He argues every AI-caused cyber breach (like OpenAI's agents hacking Hugging Face) could’ve been prevented with better sandboxing; the same tech that keeps your browser from infecting your PC. Huang even questioned why anyone would build AI if they truly believe it risks human extinction. The debate isn’t theoretical; it’s about who controls the future of AI, and whether tech giants are ready to be held accountable.