← News

ResearchSources: arXiv · 18 Nov 2026

Study of 2,664 US AI rule comments finds business commenters supply less evidence

A new arXiv paper analyzes public comments on 40 draft US AI regulations, finding that business commenters introduce less evidence than government, university, and civil society participants.

In plain words

When US regulators draft AI rules, they invite the public to submit comments, and those comments can carry evidence such as cited research. A new study analyzed thousands of these comments and found that companies cite less evidence than other groups. The authors say this complicates the idea that public comments alone let industry capture the process.

What was announced

The paper, posted to arXiv, examines how public commenters supply evidence to potential AI regulation in the USA. The authors built a corpus of 40 potential regulations and 2,664 comments from a range of stakeholders, then used computational processing to extract evidence in the form of formal citations and informal references. They report that, holding comment length constant, business commenters introduce altogether less evidence, while government, university, and civil society commenters provide significantly more research and legal sources.

The study also found that cited research is nearly evenly split between preprints and peer-reviewed articles, with both introduced by commenters of all types. Self-citation appeared across every group, but a small cluster of organizations accounted for over half of all self-citations in the corpus. The authors frame commenters as playing either an Issue Advocate or an Honest Broker role, and conclude that the findings complicate a narrative that public comments alone are a means to information capture.

What it does not show

  • The evidence is limited to the paper's abstract, so the full article's methods, data, and results are not available for verification.
  • The findings describe the 40 regulations and 2,664 comments in the corpus and do not establish how regulators actually used the submitted evidence.

Sources

Source-support review completed 19 Nov 2026. This story was drafted with AI assistance and checked against retained source passages. Automated checking is not independent verification.

Coverage: arXiv: abstract.

Updates and corrections

No corrections so far.

First published 19 Nov 2026

New to the subject? AI from the Ground Up starts from the beginning.