What Regulatory Capture Looks Like

Alex Tabarrok argues that labeling AI safety advocacy as regulatory capture misunderstands both the evidence and the theory. He notes that prominent figures - Dario Amodei, Sam Altman, Elon Musk and researchers such as Geoffrey Hinton - have warned about catastrophic AI risks long before they headed companies, and Hinton even left Google to speak freely. Calls to slow development and install independent safety teams are presented as sincere positions, not clever marketing. Tabarrok acknowledges that self-interest can influence proposals, but rejects the caricature that profit maximization explains these warnings; he insists the simplest explanation is that these actors truly believe the risks they articulate.

Tabarrok reviews classic regulatory capture theory, drawing on Marver Bernstein’s lifecycle (Gestation, Youth, Maturity, Old Age) and examples like thalidomide’s role in creating the modern FDA and the ICC’s eventual protection of railroads. He contrasts capture, which is a slow, behind‑the‑scenes process, with the current AI political moment, which is in gestation with high public salience and weaker business power per Kathleen Thelen/Culpepper-style arguments. He distinguishes ordinary rent‑seeking (e.g., Philip Morris and the 2009 Tobacco Control Act) from capture, argues regulation would reduce potential AI profits, and concludes that branding safety advocates as captors requires convoluted reasoning when the straightforward reading - that they mean what they say - is far more plausible.