Who Gets to Define the Rules for AI?

The author contends that the current trend of a handful of dominant AI labs seeking to define safety standards and control the pace of AI development risks creating a cartel-like structure that limits competition and entrenching existing market leaders. He cites the 1975 SEC bond rating system and the 1985 European Motor Vehicle Block Exemption as historical precedents where safety concerns led to monopolistic outcomes. The article critiques Anthropic CEO Dario Amodei’s recent roadmap that asks governments for antitrust waivers to allow a small group of powerful labs to agree on shared standards and limits on technology advancement. The proposal would require all other AI developers to follow these standards without public comment or consultation, effectively giving the dominant labs a legal shield to preserve their commercial advantage. The author argues that such a safety regime would only rigorously assess risks that the current leaders have built systems for, while ignoring other risks, and would set entry requirements that favor large compute operators. He maintains that safety rules should reduce risk without regard to market dominance, and that the current proposal would instead entrench the status quo. The piece calls for broader participation and transparent, evidence‑based rule‑making to avoid repeating past mistakes where safety was used to justify monopolistic control.