Major AI companies are increasingly shaping artificial intelligence laws and oversight systems in ways that may prioritize corporate interests over public accountability. Researchers from institutions including the University of Edinburgh, Trinity College Dublin, TU Delft, and Carnegie Mellon University analyzed news coverage surrounding major global AI policy events between 2023 and 2025 and identified 27 recurring patterns of what they describe as “corporate capture.”
One of the most common tactics identified in the research was “narrative capture,” where companies frame regulation as harmful to innovation or national competitiveness. According to the study, large AI firms frequently promote the idea that rules and oversight create unnecessary “red tape,” encouraging policymakers to weaken or delay regulation. Researchers also pointed to lobbying efforts, political donations, revolving-door relationships between regulators and industry, and pressure against whistleblowers and critics as examples of how Big AI companies influence governance debates.
The study compares the AI industry’s behavior to historical strategies used by industries such as Big Tobacco, Big Pharma, and Big Oil to shape public opinion and resist regulation. Researchers warn that AI governance is especially important because these systems increasingly affect healthcare, education, labor markets, public information, privacy, and democratic institutions. Some experts cited in the report argue that weak oversight could allow AI systems to spread bias, misinformation, surveillance, and concentrated corporate power at massive scale.
The findings arrive during a period of growing global disagreement over how AI should be regulated. While Europe continues pursuing stronger risk-based rules through measures like the EU AI Act, other countries emphasize innovation-first approaches with lighter restrictions. Researchers say the broader challenge is ensuring governments develop enough technical expertise and independent oversight capacity to regulate powerful AI systems effectively rather than becoming overly dependent on the companies building them.
