Big AI's Regulatory Capture Exposed: 27 Mechanisms in New FAccT 2026 Study

A new paper from Abeba Birhane and six co-authors, accepted at FAccT 2026, systematically maps how Big AI companies and governments collude to capture AI regulation. The researchers used design science research and a scoping review of literature and media reports to build a taxonomy of 27 distinct capture mechanisms across five categories: Discourse & Epistemic Influence, Elusion of Law, Structural & Institutional Influence, Economic & Resource-Based Capture, and Direct Political Influence. They then manually annotated 100 news articles, finding 249 instances of these mechanisms in action. The most common categories were Discourse & Epistemic Influence (framing narratives) and Elusion of Law (violations of antitrust, privacy, copyright, and labor laws).

Beyond quantifying the scope, the paper identifies the dominant narratives used to rationalize capture: 'regulation stifles innovation,' 'red tape,' and 'national interest' appeared most frequently. The authors argue that the coalescing forces of Big AI and governments represent a regulatory emergency that demands immediate attention from policymakers and the public. They draw lessons from historical fights against industry capture in tobacco, fossil fuels, and finance, offering transferable tactics for exposing and resisting Big AI's influence. The study also calls for envisioning counter-narratives that prioritize public interest over corporate power, urging a shift from reactive to proactive regulation.