Research Integrity

A 288-researcher randomized trial published in PNAS gave 103 teams three levels of AI autonomy in verifying published social science. AI-led teams reproduced only 37% of results, but the quieter finding sits in the middle arm: AI-assisted teams matched human-only teams on reproduction while finding fewer than half as many major errors, a verification loss with no visible signal in the finished work.

In April 2026 Nature published a multi-analyst study on an unusual scale: 457 independent analysts producing 504 reanalyses of 100 social and behavioural science studies, each given the original data and the original question and left to choose their own method. The figure that travelled was 34%, the share of claims where every reanalyst agreed with the original. The same paper reports 39% at a slightly looser bar and 80% at a simple majority, and plots the whole curve. That is the finding worth carrying. Robustness is not a number the data hands you, and a study about how a single analytic choice can carry a conclusion turns out to contain exactly that choice, disclosed openly by its authors. We read every figure off the paper itself, including what it does not establish and why preregistration will not fix it.

A Lancet-published audit of 2.5 million biomedical papers found fabricated references rising twelvefold in two years. Named methodologists immediately flagged real limits in how the finding was produced and what it can and cannot support.

UK Research and Innovation tells assessors they will never be asked to score the individual modules of a narrative CV. That is a deliberate and defensible design choice, and it also removes the audit trail that made the case against the h-index possible in the first place. Research assessment has moved from a bad number to a judgment nobody has yet measured at scale, and the first careful measurements, presented in late 2025, are ambiguous rather than reassuring.

Over fifteen years, open-science reforms became mandatory: data-availability statements, reporting checklists, preregistration badges. A 2025 scoping review in Royal Society Open Science gathered every study that had actually tested whether these interventions improve reproducibility, and found the field has mostly measured the wrong thing. Of 105 studies, just 15 looked at a direct outcome, 89 measured a proxy such as whether data were shared or a checklist completed, and only eleven asked whether results actually reproduced. The interventions may work, but after all this time we largely do not know, because we have been counting compliance instead of results.

A "Preregistered" badge tells a reader a study was run as planned. A decade of audits says otherwise: when researchers compare badged papers against their own registrations, almost all contain deviations the paper never disclosed, from a landmark Psychological Science audit where 24 of 27 studies had undisclosed departures to a 2025 study in which every one of 100 preregistered meta-analyses did. The failure is not deviation, which is often justified, but silence about it, which quietly reintroduces the exact confusion between confirmatory and exploratory results that preregistration exists to prevent. The mechanism that actually works is not more registering but someone whose job is to read the registration against the paper.

In a single reviewing season, overlapping communities of AI researchers punished reviewers for using language models, permitted other reviewers to use them, and commissioned a machine to review every submission. Each venue was internally coherent, and the ICML chairs were admirably candid that their enforcement said nothing about whether the removed reviews were any good. That candour points at the real gap: peer review has become precise at identifying who produced a review and remains unable to say whether the review was worth having. A 2008 randomised trial at the BMJ, in which reviewers found an average of 2.58 of nine deliberately planted major errors, suggests the second question was going unanswered long before any machine was involved.

Two unrelated stories from 2026, a citation-fabrication crisis and a fabricated stroke-detection dataset, describe the same failure at different layers of the research stack.

Elsevier's journal Ceramics International has retracted 56 papers in 2026, most of them after outside scrutiny rather than internal checks, Retraction Watch reported on 20 August. More than 50 were requested by an independent publishing-ethics adviser, and 43 of the notices cite comments on PubPeer. A scientific sleuth had flagged 80 of the papers to the publisher in April 2025, more than a year before the retractions began.

On 7 August the US short-selling firm Gotham City Research alleged image duplication across ten papers connected to Shinya Yamanaka, who shared the 2012 Nobel Prize for induced pluripotent stem cells. Kyoto University's CiRA has now responded, saying the report bundles cases of very different status into a single pattern claim, and that questioning Yamanaka's integrity on that basis is "completely unacceptable". Gotham's founder confirmed to Retraction Watch that the firm is short both companies whose share price the report targets. The evidence in a duplication claim can be assessed on its own terms; what a financial position changes is what gets selected for scrutiny and how it is framed.

New York University Press has pulled a 2025 monograph from sale and is destroying its remaining inventory because of a single quoted sentence. The interesting part is not that the quotation was wrong. It is that nobody invented it. The words started as a court brief's summary of a journal article, were adopted into a judge's ruling in italics with the note "emphasis added", and were then read as verbatim testimony and placed inside quotation marks by an author working from the ruling. Three defensible steps, no fabrication at any one of them, and a fake quotation at the end.

Elsevier's journal Ceramics International has retracted 56 papers in 2026, most of them after outside scrutiny rather than internal checks, Retraction Watch reported on 20 August. More than 50 were requested by an independent publishing-ethics adviser, and 43 of the notices cite comments on PubPeer.

A metascientist at Northwestern University scanned more than 270,000 antibody catalogue images and flagged over 18,000 as apparently altered, across 15 commercial suppliers. The most telling detail is not the total but the backgrounds: some 13,800 of the flagged images share just seven of them, a pattern one image-integrity specialist says implies the antibodies were never tested in a laboratory at all. The finding moves research-integrity scrutiny one step upstream of the published paper, into the catalogue a researcher reads before the experiment exists.

Journals routinely ask authors to certify they had full access to all the data in a study. Three researchers argued in Retraction Watch on 25 August that for the thousands of papers built on the TriNetX dashboard, that certification cannot be honoured: coding, cleaning and analysis decisions are made by the vendor and cannot be inspected by authors, reviewers or readers. Their proposed remedies run from a new disclosure statement to retraction where access was misrepresented.

A draft South African law would require 'practising natural scientists' to register with the state-backed SACNASP or face up to a year in prison. Two senior researchers told Nature they learned of it only after the consultation window closed. The principle of licensing scientific advice is arguable; the text's ambiguity about who counts as practising, and a custodial penalty untethered from any evidence of harm, are the parts other countries should read carefully.

All seven editors of Games and Economic Behavior resigned by 14 August after Elsevier declined to renew editor in chief Herve Moulin's contract, gave "closer alignment with Elsevier strategic priorities" as the reason, dismissed two requests for further explanation by asserting its contractual right, and began recruiting replacements without consulting the board. Fifty-eight of the journal's 62 advisory editors have gone too. It is the eighth mass editorial resignation Retraction Watch has logged in 2026, against seven in all of 2025, and at least the ninth on that list triggered by a publisher removing or replacing editors.

A preprint posted on 11 August estimates that about 89 percent of biomedical papers published in December 2025 and archived in PubMed Central carry vocabulary signatures of large language model assistance, against 77 percent across all of 2025 and 52 percent in 2024. The signal is far stronger in discussion sections than in methods, and it measures assisted writing, not authorship, fabrication or misconduct.

Elsevier's journal Ceramics International has retracted 56 papers in 2026, most of them after outside scrutiny rather than internal checks, Retraction Watch reported on 20 August. More than 50 were requested by an independent publishing-ethics adviser, and 43 of the notices cite comments on PubPeer. A scientific sleuth had flagged 80 of the papers to the publisher in April 2025, more than a year before the retractions began.

The Journal of the New Zealand College of Clinical Psychologists removed a peer-reviewed viewpoint article roughly ten months after publishing it, after the college's council decided the piece could perpetuate harm to Maori. There is no retraction notice and the article is no longer accessible, a step publication-ethics guidance reserves for rare cases such as legal orders or serious risk to health. The removal has prompted resignations and criticism from within the field, and raises a question about what happens to the scholarly record when an article simply disappears.

Sage retracted 43 papers from the Journal of Intelligent and Fuzzy Systems in July and 34 from another former IOS Press title in August, all for peer-review manipulation. Several named authors are contesting the retractions publicly, saying they were never shown article-specific evidence. The case exposes a hard question in inherited mass retractions: batch-level suspicion is strong, but the records needed to judge each paper often sit with the previous owner.

On August 1 OpenAI said an internal version of its next model, Astra, had produced proofs on ten long-standing open problems in mathematics and theoretical computer science, for about $2,000 in compute. Within days, named mathematicians told Scientific American that two of the headline results reuse earlier published arguments without citing them, and that the release invoked a declaration on AI in mathematics while bypassing the peer review that would have caught the omission. The dispute is a useful early template for how an 'AI solved X' claim should be read.

404 Media reported that Research Gold, which sells systematic reviews and meta-analyses to medical researchers on an explicit promise of no AI involvement, staffed its About page with people who do not exist and lifted the identities of real methodologists from LinkedIn without permission. It quoted $1,900 for a review of a deliberately nonsensical question. The disclosure claim was the product being sold.

An updated preprint by Anna Abalkina and colleagues matched 2,062 papers in 366 IEEE conference proceedings to publication offers advertised on social media, and Science reports that 66 have been retracted so far. The count rose from 1,720 in the April version without any change in method, which tells you the figure is a floor rather than an estimate. The wider point for anyone screening literature is that conference proceedings sit outside every journal-level check a researcher would normally apply.

Anthropic has begun embedding an invisible watermark in text generated by Claude and attaching signed C2PA provenance metadata to the files it produces, in response to the EU AI Act's transparency code. The company's own documentation is candid that a detected mark only shows content was processed by Claude, not written by it, and that an absent mark proves nothing. The tools that would let anyone outside Anthropic check for the mark have not been published yet.

Elsevier's International Journal of Biological Macromolecules has retracted 120 papers since June for systematic peer review manipulation in guest-edited special issues, nearly all published in 2024 and 2025 while the journal's annual output doubled. The count is the least useful figure in the story. One researcher reported 352 suspect papers to the journal and 12 have been retracted, and the papers already sit in reference lists a notice cannot reach.

JAMA Internal Medicine has declined for the third time to retract a 2000 study concluding that controlled-release oxycodone rarely caused withdrawal in osteoarthritis patients, after clinicians pointed to 2007 court records showing the sponsor, Purdue Frederick, withheld contradicting withdrawal data. The journal gave two reasons: no evidence of fabrication or falsification, and no agreement from all surviving coauthors. Both are narrower than what COPE and the ICMJE actually ask of editors, and the gap is the story.

The American Journal of Political Science published a revised AI policy for authors and reviewers on 11 August 2026. Alongside the familiar requirements, no AI authorship and disclosure of any use that shaped the research, it asks for something most journal policies do not: the degree of human oversight applied to AI output, meaning whether it was reviewed, edited, or approved before it entered the manuscript. Non disclosure is stated grounds for decline. The revision lands while COPE and four partner organisations work toward a global AI disclosure standard due at the end of 2026.

A widely used Kaggle dataset marketed for stroke-detection research contained celebrity photos and images lifted from a dental journal. Kaggle acted only after a copyright complaint, not the original research-integrity report.

AJPS published a new AI policy on 11 August requiring authors to name every AI tool used in a manuscript, and barring reviewers from uploading manuscripts to any AI tool at all.

Clarivate confirms Heliyon will regain its Web of Science impact factor in October, two years after a quality hold and a mass self-correction of 1,197 items.