We Can't Let Enormous Weirdos Regulate AI

This is a polemic arguing that a small, socially insular clique of well-funded AI “safety” advocates and frontier companies are steering public policy toward regulating AI in ways that serve their interests rather than addressing the most immediate, evidence-backed harms. Using a satirical drone analogy about “thumbed dogs” and eccentric Drone Safety Researchers, the piece contends that obsession with speculative existential risks has crowded out practical concerns - airline safety, job loss, scams, privacy breaches - and has sidelined experienced, conventional experts. The author highlights recent flashpoints: a high-profile resignation by Jacob Coxon claiming runaway self-improvement risk, reports of models hacking other systems (including an Australian government site), and public warnings from Anthropic’s Dario Amodei and OpenAI’s Sam Altman at the U.N.

The argument calls out concrete proposals and conflicts of interest: Amodei’s “We Must Pace the Frontier” pitch for embedded evaluators and industry-wide standards, mentions of METR’s social ties to firms it would oversee, and a footnote about possible antitrust waivers that could enable cartel-like coordination. The critique stresses that the doom contingent simultaneously pushes for tighter control while continuing to build capabilities and accruing massive wealth, risking capture of regulation, reduced competition, weakened democracy, and the marginalization of more pragmatic safety work unless the clique’s influence is broken.