Scientific Reports is under PubMed review as researchers debate AI’s role in papers, peer review and research integrity.


Research integrity was the week’s connective thread: a major journal is being reassessed for PubMed indexing, a publisher pulled 1,400 conference papers in a single day, and researchers are testing what AI can and can’t be trusted to do.
Retractions and journal scrutiny
Retraction Watch reported that a new University of Sydney investigation of astrophysicist Joachim Schmidt led to retractions. The Game Theory Society withdrew its support for a journal after Elsevier did not renew its editor-in-chief’s contract. A federal judge dismissed a $1 billion defamation suit against Elsevier, while a Springer Nature journal in Norway could lose its registry standing after clearing an author of duplication.
The ruling in the Elsevier case and the journal’s handling of the duplication allegation point to different pressures on publishing institutions. Meanwhile, Scientific Reports is on hold in PubMed while the U.S. National Library of Medicine reevaluates the journal. That review puts a large publisher’s standards and indexing status under close attention.
The scale of integrity monitoring continues to grow. The Hijacked Journal Checker now lists more than 450 entries, and the Retraction Watch Database contains more than 68,000 retractions. Retraction Watch’s COVID-19 retractions list is nearing 800 entries, while its mass resignations list has passed 50. The organization invites readers to support its work with a tax-deductible donation.
AI puts research practices to the test
Several reports and studies this week examined AI’s place in scientific work. An IEEE Spectrum writer described using AI to fake a scientific paper, while researchers and commentators asked whether AI-generated mathematical proofs can be trusted. A preprint assessing 500,000 science, engineering and computing papers follows an earlier analysis that found fabricated references in biomedical literature in one in 277 PubMed-indexed papers.
The debate has moved into peer review itself. Researchers have laid out the potential and pitfalls of AI-assisted review, as Springer Nature and Google DeepMind work on an AI peer-review assistant. Other perspectives ask whether AI could become the better referee and consider giving AI a role in peer review. A preprint reports a higher share of possible AI-generated abstracts in fully open-access medical journals than elsewhere. Another essay revisits open science in the age of AI.
Questions about authorship and accountability extend beyond journals. Dartmouth’s provost resigned after his work was flagged as AI-written. A study of management research reports dominance by a scientific elite, a pattern Retraction Watch has also explored in business-journal networks. A separate analysis finds serious weaknesses in research on AI in education: its audit of 39 meta-analyses found 39 positive results.
Methods, incentives and accountability
Other work focused on how research is produced and assessed. A preprint suggests three warning signs that may help identify questionable journals, drawing on a list of red flags shared by its authors. A discussion of editorial decisions asks whether accountability belongs to a clique or the journal community. Another paper considers whistleblowing and retractions in Iran-affiliated medical research, including motives and editorial perspectives.
Researchers are also revisiting familiar claims and tools. Sholto David’s search for papers with erroneous claims about protein expression focuses on reported protein expression that should be absent. A critique of the Stanford prison experiment’s lasting hold revisits the influence of controversial psychology studies. And a Dutch journalist challenged an article published in the American Journal of Political Science.
Proposals for reform include making Registered Reports the default route for confirmatory studies and compensating peer reviewers instead of treating their labor as compulsory. A European preprint-centered publishing model has received a major boost. Meanwhile, a post examines self-confessed fabrications and the rise of TBDs, and a commentary argues that researchers who lie about their work undermine their place in the scientific community.
The week also brought reports on the American Dental Association, which said a “threat” preceded members’ ejections but acknowledged its policies were unclear, and on the planned reopening of UK Biobank to researchers after a data breach. OpenAlex has introduced trusted journal lists. Ecology researchers are weighing the promise of “Big Code” alongside the errors it can conceal, while a paper makes the case for careful bibliography practices. A planned protest at Oxford University centers on an academic who accused Jason Arday of plagiarism.
Upcoming talk
Retraction Watch’s Ivan Oransky will take part in the virtual 2026 Council of Science Editors Fall Virtual Symposium session, “Rethinking Incentives to Publish: Aligning Accountability Across Stakeholders,” on November 19. The week’s stories suggest why that question keeps resurfacing: research integrity depends on incentives and scrutiny across the whole publishing system.
Like Retraction Watch? You can make a tax-deductible contribution to support our work, follow us on X or Bluesky, like us on Facebook, follow us on LinkedIn, add us to your RSS reader, or subscribe to our daily digest. If you find a retraction that’s not in our database, you can let us know here. For comments or feedback, email us at [email protected].
Like Retraction Watch? You can make a tax-deductible contribution to support our work, follow us on X or Bluesky, like us on Facebook, follow us on LinkedIn, add us to your RSS reader, or subscribe to our daily digest. If you find a retraction that’s not in our database, you can let us know here. For comments or feedback, email us at [email protected].
Discussion
Sign in to join the discussion.