Senator Bernie Sanders introduced legislation this week that would ban the development of artificial superintelligence and imprison violators. Simultaneously, a New Yorker documentary titled 'Josh's Wedding' captures an OpenAI engineer and his groomsmen debating the future of the technology one of them is actively building. The juxtaposition is almost too clean: the legislative branch reaching for the crudest possible instrument while the people inside the machine conduct their own informal ethics committee at the open bar.
AI Regulation and the Regulation Gap
A 2026 report covered by TechCrunch found that even daily AI users are worried about the technology, and that greater exposure does not reduce anxiety or opposition to regulation. This matters because it dismantles the standard industry counter-argument: that skepticism is just ignorance, and familiarity will cure it. The groomsmen in the New Yorker film aren't uninformed. They work in the field. And yet. Fast Company reports that researchers are articulating four distinct doomsday scenarios, from weaponized biodesign to autonomous economic manipulation. Meanwhile, Fast Company also notes that while the US argues about AI's name, China is writing the actual rules.
The Regulation Imagination Problem
Sanders' bill is almost certainly unenforceable. You cannot define superintelligence precisely enough to write a criminal statute around it. But the bill's bluntness is also diagnostic: the legislative imagination for AI regulation is so underdeveloped that criminalization of a vague technical threshold is the best available option. The OpenAI engineer's wedding, meanwhile, is doing the more interesting work. Informal deliberation among people with actual technical knowledge is where the real ethics policy gets formed, or fails to. The question isn't whether Sanders' bill passes. The question is what fills the vacuum when it doesn't.