AI Governance Has No Center, and Everyone Notices a Different Hole

A student on Bluesky noted this week that her school now requires a mandatory discussion about AI pros and cons in every class. The detail that stuck with her wasn't the policy — it was the consensus argument her peers keep returning to. By far, she wrote, the most-cited problem is the water analogy: the idea that AI is just a tool, like electricity or running water, and therefore regulation should be minimal and light-touch. She found this remarkable not because it was wrong but because it was everywhere, repeated with the confidence of someone who had arrived at the thought independently.

That kind of distributed, uncoordinated convergence on a single frame is itself a form of regulatory pressure — and it's happening at the same moment the institutional machinery is visibly fragmenting. The legal and policy coverage this week reads like a map of a country that has decided to regulate AI in every direction at once. The U.S. Equal Employment Opportunity Commission released new technical guidance on employer use of AI and disparate impact. California is contemplating separate AI employment rules. Colorado is building what one law publication described as a "partnership model" between AI deployers and developers. Ontario is pushing insurers to justify automated decisions. The House of Lords weighed in on automated decision-making in the UK public sector. The EU's AI Act is drawing criticism from Human Rights Watch for endangering social safety nets. None of these efforts are talking to each other.

One Bluesky commenter put the structural problem plainly: AI doesn't fit neatly into existing political narratives about government overreach or market failure, which may explain the absence of clear policy frameworks. That framing has been circulating in Canadian political commentary following Pierre Poilievre's appearance on a podcast, where the absence of a coherent conservative — or liberal — position on AI governance became the subtext of every exchange. The observation applies just as well south of the border. A post on Bluesky linking to a Tech Policy Press podcast described researchers who have started treating AI hype itself as an object of study — calling it "Hype Studies" — and trying to understand the social and political dimensions of overpromising before anyone has agreed on what the technology should be allowed to do. That the study of hype has become a research discipline is a sign of how far the gap between rhetoric and governance has widened.

The most anxious voices in the conversation aren't opposed to AI — they're opposed to who they expect will govern it. One Bluesky post noted, with undisguised alarm, that Elon Musk had given lectures in Rome framing AI regulation as "the antichrist," and that this same person is positioned to influence how the U.S. federal government approaches the technology. Another post, more measured but reaching the same conclusion, called "AI-centered governance" frightening, particularly under assumptions of what the author called "the right's unreality." These aren't fringe positions — they're the framing that keeps reappearing in the most-engaged posts on the beat this week. The fear isn't anarchy; it's captured governance.

What the self-described anti-AI commenter said — the one who nonetheless called clear lab policies "a valid starting point" — captures where the most pragmatic part of the conversation has landed: not demanding perfect regulation, but demanding legibility. Tell people what the rules are, even if the rule is "please don't." That request sounds modest. Against the backdrop of a patchwork of state-level employment laws, a fragmented EU framework, a federal government without a coherent position, and a classroom where every student has independently concluded that water flows downhill and so should AI — it's actually a significant ask. The governance conversation isn't converging on a model. It's converging on the recognition that no model is coming fast enough to matter.