Weekend reads: AI slop crackdown, baby's poisoning questions, and retractions sparked by social media

Weekend reads: research integrity, AI clampdowns, and what PR and research teams should do next

If your week blurred past, here's the fast catch-up. Big moves in retractions, AI policy shifts, and a string of cautionary tales for anyone managing scientific reputation or research output.

This week's standout stories

• Study theft saga: a stolen project was sold and published - and the original researcher then faced a plagiarism claim.
• Plagiarism flags hit an engineering journal after a poultry paper was pulled.
• A medical journal ran a letter on AI with a fake reference citing the journal itself.
• Nearly 150 papers retracted for compromised peer review.
• Satire meets scrutiny: a fake "pregnancy cravings for prime numbers" paper got published - on purpose - to expose weak checks.

By the numbers

Signal, not noise: the hijacked journal tracker now lists 400+ titles. A retraction database has over 63,000 entries, including 640+ tied to COVID-19, and a mass-resignations tally has reached 50. The pattern is clear: integrity issues are frequent, global, and PR-sensitive.

What else moved the needle

• Accountability: A report asks, "Did a celebrated researcher obscure a baby's poisoning?" Questions linger, and so will public interest.
• Social media pressure: Analyses link critical posts to later retractions - public critique is shaping outcomes.
• Preprint guardrails: arXiv tightened submissions, requiring first-timers to get endorsements and insisting on English-language submissions. See policy details on arXiv endorsements.
• AI content quality: Multiple pieces argue that "AI slop" is overwhelming screening and peer review.
• Bibliometrics: Five major challenges in medical metrics highlight how incentives can distort impact.
• Replication debate: Expect more push-pull over what counts as a valuable replication and how it gets credited.
• Automation: The UK government is backing AI that can run lab experiments - watch the regulatory and safety angle. Early signals are in government releases.
• Misconduct and law: Where "fudging" ends and fraud begins - legal exposure is getting clearer.
• Robots in the lab: A "robot chemist" paper was corrected; open questions remain about claims and oversight.
• Fake references: An ethics journal retracted a whistleblowing paper with nonexistent citations.
• Image manipulation: A deputy department chair lost another paper over figures - repeat patterns matter in crisis planning.
• Rankings gaming: Concerns that some private universities are gaming systems will keep PR and admissions teams busy.
• OpenAI for science: Leadership interviews hint at where AI-enabled discovery is heading - and the comms scrutiny it will bring.
• Peer review ideas: Calls to bring more practitioners into review to test real-world relevance.
• Environmental research ethics: Mapping common ethical pitfalls in methods and data.
• Generative tools risk: One researcher lost two years of work after toggling off ChatGPT data consent - tooling policies matter.
• Faster publishing: COVID-era practices proved useful; some argue to keep them.
• Authorship order: Name order can choke collaboration - expect more transparency policies.
• Gender studies scrutiny: Why some results collapse under review - methodology and bias in the spotlight.
• Deals and access: Three more UK universities stepped back from new big-publisher contracts.
• Clinical trials: Guinea-Bissau suspended a US-funded vaccine trial amid concerns from regional scientists.
• Puberty blockers: Growing calls from medics, lawyers, and the public to pause a controversial trial.
• Systematic review bias: Early signs of "reverse spin bias" in medical reviews.
• Training culture: Warnings about worsening ethics in biomedical research and weak mentorship.
• Global retractions: Vietnam's retraction rates are among the world's highest; collaborations with Saudi Arabia face their own retraction problems.
• Genomic data risk: Genetic data from 20,000+ US children was misused for "race science" - renewed calls to safeguard datasets.
• Forensic oversight: A biologist exposed a DNA lab scandal in Australia - chain-of-custody and validation stay critical.
• Peer review tone: "Linguistic snobbery" in reviewer comments can harm early-career researchers.

Why this matters for PR and research leaders

• Tighten source checks: Require authors to verify every reference and figure. Spot-audit before submission and after acceptance.
• AI usage policy: Set a written policy for AI in writing, translation, and analysis. Require human editing for language quality and disclosures for any AI assistance.
• Preprint readiness: If posting to arXiv, line up endorsers and ensure English clarity before upload to avoid delays or rejections.
• Social listening: Track critical posts about your work. If legitimate, respond with data, corrections, or an investigation timeline.
• Retraction playbook: Have a play-by-play for allegations - intake, fact-finding, external review, statements, and updates. No slow-rolling.
• Image and data forensics: Use tools for duplication detection and stats anomalies. Train teams to spot common artifacts.
• Authorship and contributions: Move beyond name order; publish contribution statements and conflict disclosures.
• Data governance: Lock down access, consent, and reuse policies for genomic and sensitive data. Pre-approve external collaborators and audits.
• Peer review diversity: Add practitioners to review pools for translational work; they catch practical flaws early.
• Educate and upskill: Give staff training on AI quality control and reference hygiene to avoid "AI slop" creeping into submissions.

Upcoming talks

• Maintaining Integrity in Peer-Reviewed Publications - Jefferson Anesthesia Conference 2026, featuring Adam Marcus (February 2, Big Sky, Montana)
• Responding to Research Misconduct Allegations - AAAS EurekAlert! webinar featuring Ivan Oransky (February 3, virtual)
• Scientific Integrity Challenged by New Editorial Practices - featuring Ivan Oransky (February 12, virtual)

Helpful resources

• arXiv endorsement policy for first-time submitters: info.arxiv.org/help/endorsement
• If you're setting team standards for AI-assisted writing and review, these practical courses can help: Latest AI courses

If you spot suspect work or a potential retraction, escalate early: document evidence, notify the journal with specifics, and brief leadership on timelines and risks. Silence makes problems bigger; transparency builds trust.

Posted on January 31, 2026