Artificial intelligence (AI) companies influence policy and regulation using methods similar to those employed by the tobacco, pharmaceutical, and oil industries. A new study mapped the influence techniques of "Big AI" companies. These are firms that develop and deploy large-scale AI technologies.
Researchers from the University of Edinburgh, Trinity College Dublin, TU Delft, and Carnegie Mellon University analyzed news articles. They sought evidence of methods used to control narratives and influence AI-related policy measures. The team identified 27 established patterns of "corporate capture." This is a process where regulations and public bodies serve corporate interests over public interests.
The researchers examined 100 news stories. These stories were published around four global AI events between 2023 and 2025. The events included the European Union (EU) AI Act trilogues and global AI summits in the United Kingdom, South Korea, and France. They found 249 instances fitting corporate capture patterns.
"Narrative capture" was a prevalent mechanism. This involves attempts to influence the positions or decisions of public officials and regulations. Dominant narratives included claims that "regulation stifles innovation" and creates "red tape." This portrays regulation as unnecessary or excessive, leading to calls for deregulation. Another common tactic was "elusion of law," involving violations and contentious interpretations of antitrust, privacy, copyright, and labor laws.
Big AI has also resisted oversight through lobbying and retaliation against whistleblowers, researchers, and lawmakers. The industry has benefited from a "revolving door" model. Former policymakers often advise or join major AI companies. Additionally, there are instances of significant donations to political parties and public officials holding equity in regulated companies.
Experts suggest learning from other industries to prevent corporate capture. This includes separating public and private interests. Binding rules for government-industry interactions could manage conflicts of interest. The findings will be presented at the Association for Computing Machinery (ACM) Conference on Fairness, Accountability, and Transparency in June 2026.
Related stories
New LED Technology Powers Insulating Nanoparticles for Infrared Light
Scientists developed a new LED technology that powers insulating nanoparticles with organic molecular antennas, producing ultra-pure near-infrared light for medical imaging and communication.
CAR T-Cell Therapy Shows Promise for Autoimmune Diseases
CAR T-cell therapy, a cancer treatment, is being explored for autoimmune diseases like multiple sclerosis, aiming to reset the immune system by targeting specific cells.
Falling Space Debris Poses Growing Risk as Spacecraft Materials Improve
Falling space debris is an increasing risk as stronger, more heat-resistant spacecraft materials prevent complete atmospheric burn-up, leading to more objects reaching Earth's surface.