Every previous turning point featured a contest in which regulatory capture was a central dynamic. The canal companies captured parliamentary oversight of waterway regulation. The railway companies captured rate-setting boards. The broadcast media captured the FCC. The platform companies of the ICT age captured regulatory processes in both the US and Europe, producing decades of permissive treatment that contributed to the incomplete deployment phase.
AI capture operates through familiar mechanisms with some novel features. The lobbying expenditures of technology companies in the United States and Europe have increased dramatically since 2023. The revolving door between regulatory agencies and AI companies continues to spin. The concentration of technical expertise — most people who deeply understand AI systems work for AI companies — means that regulators depend on industry sources for the information they need to regulate, creating structural information asymmetry that favors the regulated.
The capture risk is particularly acute for supply-side regulation because it directly affects the competitive dynamics of the AI industry. Demand-side regulation — investment in citizen capability, education, transition support — attracts less intense lobbying because it does not directly affect industry profitability. This asymmetry is one of the reasons demand-side regulation is both more resistant to capture and more consequential for the deployment phase.
Resisting capture requires institutional mechanisms that have historically proven effective: public funding of independent technical expertise, robust whistleblower protections, transparency requirements for regulator-industry interactions, and — most fundamentally — strong political coalitions that can counterbalance incumbent interests during the turning point's window of opportunity.
The concept of regulatory capture was developed in the public choice literature by George Stigler and others in the late twentieth century. Perez's framework integrates it into the broader pattern of installation-phase incumbents shaping turning-point outcomes.
Structural dynamic. Parties most affected by regulation invest most in shaping it.
Multiple mechanisms. Lobbying, revolving door, technical expertise asymmetry, campaign contributions.
Historical pattern. Every previous turning point featured capture contests.
AI-specific features. Technical expertise concentration amplifies information asymmetry.
Countermeasures. Independent expertise, transparency, whistleblower protections, strong reform coalitions.