Northwestern and NIH researchers urge rethinking plagiarism rules for AI-assisted research writing

A new commentary by researchers at Northwestern University and the National Institutes of Health argues that plagiarism rules need to change as generative AI becomes more common in research writing.

Published May 18 in Nature Machine Intelligence, the piece says existing plagiarism standards are built mainly around verbatim copying, but AI tools can now paraphrase text or reproduce ideas in ways that are harder to spot.

The authors, David Resnik of NIH and Mohammad Hosseini of Northwestern, say research-integrity policies should shift toward clearer rules on plagiarism of ideas in AI-assisted writing.

Why the authors want a rethink

Northwestern's summary of the commentary says the authors see a gap between current misconduct rules and how scientists increasingly use generative AI to draft and revise text.

Their concern is not just copied wording. It is also the possibility that AI systems can restate another person's ideas closely enough to obscure the original source while still avoiding direct text-match detection.

The commentary argues that research-misconduct definitions should explicitly address generative AI use so institutions and journals have clearer standards to enforce.

Why it matters

NIH policy pages still describe plagiarism as a form of research misconduct and warn researchers to avoid it, including when using AI tools.

That leaves journals, universities, and funders with a practical problem: policies that were written for traditional copy-and-paste misconduct may not fully capture AI-assisted paraphrasing or idea borrowing.

The article does not say any policy has changed yet. The immediate development is the publication of the commentary and the call for updated standards.

What happens next

For now, the proposal is a policy argument rather than a formal rule change. Whether journals or research-integrity bodies adopt the language shift remains open.

The authors also want clearer guidance from institutions on how to define and police plagiarism when generative AI is part of the writing process.