AI-Generated Language Is Beginning to Pollute Scientific Literature https://t.co/yxNB4LEGvL #AInews #artificialintelligence #AI #artificialintelligencenews

— AI & Robotics News - Unite.AI (@UniteAi) July 15, 2021

‘Tortured phrases’, lost in translation: Sleuths find even more problems at journal that just flagged 400 papers https://t.co/jTTJMyoVIg

— Retraction Watch (@RetractionWatch) July 19, 2021

Tortured phrases: A dubious writing style emerging in science. Evidence of critical issues affecting established journals https://t.co/iZU7ZDKa7d [Led to this by Jeffrey Aronson’s post at https://t.co/geaZEi7Xql]

— Tom Roper (@tomroper) July 31, 2021

Analysis reveals that strange turns of phrase may indicate foul play in science. https://t.co/Kmw7nd20Nn

— nature (@Nature) August 5, 2021

Jeffrey Aronson: When I Use a Word . . . Artificial translation - The BMJ https://t.co/Qoqm4JsBn6 #ai #ml #dl

— AI News Feed (@AINewsFeed) August 7, 2021

“Tortured” phrases used to fool plagiarism detectors now infest scientific papers https://t.co/ndCKRyd0v8

— Boing Boing (@BoingBoing) August 9, 2021

Recent research found that hundreds of published papers included "tortured phrases" that may indicate plagiarism or fabrication. https://t.co/IfP5ExRwTN#plagiarism

— Jonathan Bailey (@plagiarismtoday) August 10, 2021

Bizarre phrases betray fake research papers, physics rules the roost in Peckham and Brentford FC in this week's #TheRedFolder https://t.co/rtHDiS6y4w

— Physics World (@PhysicsWorld) August 13, 2021

We hear a lot about misinformation being put about by non-scientists, but nothing about fraudulent papers published in scientific journals. Former Google software engineer Mike Hearn investigates for the Daily Sceptic. https://t.co/mifiR4iSLL

— The Daily Sceptic (@LD_Sceptics) August 10, 2021

Tortured phrases: “This reminds me of the time in #Friends when Joey used a thesaurus to write an adoption recommendation letter for Monica and Chandler. Except this is real and more is at stake.” 😂 Nice catch @logcface https://t.co/5JaX2PsnbV https://t.co/LVGI1vMCqR

— Guillaume Cabanac (@gcabanac) August 14, 2021

“SPIN to WIN the Battle of Publications: Recent Trends of Scientific Fraud and Misconduct in Manuscript Preparation” editorial by @Pratyus11631733 in @KBiotechexpress (Aug 2019) HT @scipapers2 https://t.co/EDcOsqlI2a related to ‘tortured phrases.’ https://t.co/H3f6Mu88Rj

— Guillaume Cabanac (@gcabanac) August 15, 2021

Scholars discover a mysterious wave of scientific papers using odd, “tortured phrases” https://t.co/KcHNHUgqvJ

— Boing Boing (@BoingBoing) August 30, 2021

#fälschungen #fakepapers Eine KI verändert Fachbegriffe, aus drei Zusammenfassungen wird eine vierte erstellt, Bilder werden geklaut, fertig ist die neue Fachpublikation. Tausende wissenschaftliche Veröffentlichungen werden so gefälscht - und publiziert. https://t.co/rU3Ou4F3so

— NZZ Wissenschaft (@NZZWissen) September 1, 2021

Hundreds of junk-science papers have been retracted from reputable journals after fraudsters used ‘special issues’ to manipulate the publication process. And the problem is growing. https://t.co/FSKKvfV2ez

— nature (@Nature) November 9, 2021

Hoe kwamen onzinbijdragen in een tijdschrift van een gerenommeerde wetenschappelijke uitgever? „Het is alsof je olie en azijn bij elkaar gooit. De delen van het artikel mengen gewoon niet.” https://t.co/5wnzOwAK9N

— NRC (@nrc) December 4, 2021

Quand l'informatique détecte des expressions scientifiques douteuses https://t.co/hxc3oaH3Gc

— Sciences et Avenir (@Sciences_Avenir) December 20, 2021

Un chercheur toulousain distingué par la revue Nature pour ses travaux sur les fausses études scientifiques https://t.co/m1kTxTIAfR pic.twitter.com/5vYRB3Ecya

— France 3 Occitanie (@F3Occitanie) December 22, 2021

Parmi les 10 personnalités qui ont marqué la science en 2021, nous avons interviewé Guillaume Cabanac, enseignant-chercheur à l’université, surnommé « détective de la tromperie » par @Nature il débusque les fausses publications scientifiques #Nature10https://t.co/hWXJSAIKTI pic.twitter.com/dcYsQ2xTMT

— UT3 Paul Sabatier (@UT3PaulSabatier) January 5, 2022

Distingué pour ses travaux par la prestigieuse revue "Nature", Guillaume Cabanac, un informaticien toulousain, piste les publications scientifiques « bidon » @UT3PaulSabatier https://t.co/E3I2bmGDnb via @20minutes pic.twitter.com/PDd5XCrSAb

— 20 Minutes Toulouse (@20minutestoul) January 5, 2022

Ce chercheur toulousain fait partie des 10 personnalités scientifiques mondiales de l'année @UT3PaulSabatier @gcabanac https://t.co/UR5VAAsqq0 via @ladepechedumidi

— Cyril Doumergue (@cyrildoumergue) January 6, 2022

À travers ses travaux et sa plateforme, @gcabanac cherche à dépolluer la littérature scientifique #science #recherche #sciences @IRIToulouse @UT3PaulSabatier https://t.co/1lDAUay7uQ

— Paul Périé (@paul_perie) January 12, 2022

Modern plagiarists are making use of software and perhaps even emerging AI technologies to draft articles—and they’re getting away with it. https://t.co/fBThIjRwnt

— Bulletin of the Atomic Scientists (@BulletinAtomic) January 13, 2022

Fraudes scientifiques : la traque s'organise https://t.co/FYVDpXxbkE

— Les Echos (@LesEchos) January 14, 2022

« Problematic Paper Screener : détection d'expressions torturées révélatrices d'articles frauduleux » interview diffusée dans le 19/20 de @F3Occitanie du 07/02/2022. Une recherche @UT3PaulSabatier @IRIToulouse. Compléments : https://t.co/OkFHIfnMgL et https://t.co/LvROSUfpoJ pic.twitter.com/nAtBvpP6Yd

— Guillaume Cabanac (@gcabanac) January 20, 2022

« Dépolluer la littérature scientifique » : mon logiciel ‘Problematic Paper Screener’ https://t.co/fSmzEq49x9 est un indic fournissant des tuyaux à des détectives scientifiques qui signalent les problèmes sur @PubPeer 🔎. Interview avec @OliviaSchaller pour @telematin @France2tv pic.twitter.com/pAkvva7eu9

— Guillaume Cabanac (@gcabanac) January 21, 2022

« Un logiciel pour traquer les fausses informations dans les publications scientifiques » interview pour @RadioPresence https://t.co/hUkWitopqe à propos du ‘Problematic Paper Screener’, une recherche @UT3PaulSabatier @IRIToulouse. Pour compléter : https://t.co/LvROSTXOxb

— Guillaume Cabanac (@gcabanac) January 25, 2022

La #science « s’auto-corrige » lorsque des découvertes publiées permettent de réviser les connaissances accumulées au fil des siècles. Vrai aussi lorsque des scientifiques identifient des #fraudes et les font supprimer 🧹. Mon interview sur @viaOccitanieTV https://t.co/YZWXCz0LpI pic.twitter.com/e4CSKdAiWC

— Guillaume Cabanac (@gcabanac) January 28, 2022

#CNRSnews 📰 | This researcher tracks “tortured” expressions in scientific publications to identify those that are unreliable. The software program that he co-developed for this purpose earned him a spot in the 2021 Top 10 selected by @Nature ⤵https://t.co/lqDEqcItU8

— CNRS 🌍 (@CNRS) March 9, 2022

Vidéo. Guillaume Cabanac, « dépollueur scientifique » https://t.co/vAIMTfeJxT

— Le Monde (@lemondefr) August 1, 2022

Guillaume Cabanac, le Sisyphe de la dépollution de la science https://t.co/YsnFkHwwa5

— Le Monde (@lemondefr) August 15, 2023

Scientific sleuths like @gcabanac are picking up dishonest use of ChatGPT in peer-reviewed papers.
Authors didn't disclose they'd used ChatGPT, but didn't delete the phrases 'As an AI language model...' or 'Regenerate response'. 🤦
By @gvconroy | @Naturehttps://t.co/WCjJC88y83

— Richard Van Noorden (@Richvn) September 8, 2023

Guillaume Cabanac traque les « fake news » dans la littérature scientifique https://t.co/LMZracm1rj

— Le Point (@LePoint) January 15, 2024

Just out @arxiv: “Tortured phrases: A dubious writing style emerging in science. Evidence of critical issues affecting established journals” w/ Labbé & Magazinov https://t.co/eh0YNO5T8v Probable systematic manipulation of the publication process of Microprocessors in Microsystems pic.twitter.com/vrVQUi3TNn

— Guillaume Cabanac (@gcabanac) July 15, 2021

I once started an academic misconduct procedure to a student that gave me a dissertation where 'carbon black' had been replaced by 'acetylene dark' and 'thin film' with 'slim movie'. https://t.co/iSJbzvOzp11

— Dr Sara Ronca (@sararonca77) August 5, 2021

🕵️‍♀️ Open Call for Investigation: Let us (re)assess the published literature featuring ‘tortured phrases’ and report problematic papers on @PubPeer. Case of “credulous/innocent/gullible Bayes” 🤪 via @DSDimensions sorted by decreasing citation counts: https://t.co/kVjBezRIGX https://t.co/H3f6Mu88Rj pic.twitter.com/OTNR0ERX6f

— Guillaume Cabanac (@gcabanac) July 18, 2021

I've seen the same tortured style in student essays when I was tutoring students!

They would actually pay services to rewrite plagiarised texts like Wikipedia, like here most of that service was automated by random word-replacement with synonyms; it rarely worked well https://t.co/BYhQE9j19S

— Philipp Bayer (@PhilippBayer) August 5, 2021

#Tortured Phrases in a 2021 @IEEEorg conference paper. Forget about ‘artificial neural networks’ and ‘random forests.’ Embrace 𝘤𝘰𝘶𝘯𝘵𝘦𝘳𝘧𝘦𝘪𝘵 𝘯𝘦𝘶𝘳𝘢𝘭 𝘰𝘳𝘨𝘢𝘯𝘪𝘻𝘢𝘵𝘪𝘰𝘯 and 𝘢𝘳𝘣𝘪𝘵𝘳𝘢𝘳𝘺 𝘣𝘢𝘤𝘬𝘸𝘰𝘰𝘥𝘴 😵 Details on @PubPeer https://t.co/wxVZaVWuEQ pic.twitter.com/nQxmCDtTlk

— Guillaume Cabanac (@gcabanac) August 4, 2021

Still working on the "tortured phrases" project after reading the preprint by @gcabanac, Labbé, and Magazinov, arXiv:2107.06751 - https://t.co/CyOZreFPxn - finding dozens of these in @ElsevierConnect 's
Microprocessors and Microsystems and in Materials Today.

— Elisabeth Bik (@MicrobiomDigest) August 10, 2021

Top 1 per citations according to @DSDimensions (https://t.co/93wGr3AgXr) is a 2018 @SpringerNature article cited 111 times. The tortured phrase appears in the ‘literature review’ section. Some authors ‘synonymise‘ the abstracts of a few related articles then copy-paste et voilà! pic.twitter.com/3b1yPbR53u

— Guillaume Cabanac (@gcabanac) August 5, 2021

As a person who uses translation software daily with Mandarin, I recognized some of these types of what they call "tortured phrases" immediately. https://t.co/3eghBQzP5A

— B. E. Happy (@happaney) August 11, 2021

“A tortured phrase claims heterosexuality of the carbon structure” by Jaime A.Teixeira da Silva in 𝘙𝘦𝘴𝘶𝘭𝘵𝘴 𝘪𝘯 𝘗𝘩𝘺𝘴𝘪𝘤𝘴 https://t.co/ipotommPjG tipped by a @PubPeer comment https://t.co/2B96XP2Kfn by M.V. Simkin #crowdsourcing HT @freddie2310 pic.twitter.com/qce9xEHFjS

— Guillaume Cabanac (@gcabanac) September 27, 2021

The @IEEEorg (left) and @SpringerNature (right) published the same text full of tortured phrases authored by the same two persons. See https://t.co/xbRw10SGJx and https://t.co/tt6BRSf3Iq pic.twitter.com/PCT5G4I6rF

— Guillaume Cabanac (@gcabanac) January 2, 2022

A 2021 @SpringerNature chapter paraphrasing @timberners_lee with synonyms and full of tortured phrases. Sold 29.94 €. Details on @PubPeer https://t.co/2p3b4PBcjq and other cases in the ‘Problematic Paper Screener’ https://t.co/fSmzEq49x9 (volunteers welcome: let's decontaminate) pic.twitter.com/yXjYRdzAuh

— Guillaume Cabanac (@gcabanac) January 30, 2022

Hundreds of nonsensical computer-generated articles, spotted years after the problem was first seen, could lead to a wave of retractions. https://t.co/u89i4315oT

— nature (@Nature) May 27, 2021

Un nouvel outil pour traquer les articles scientifiques « bidons » écrits par des logiciels https://t.co/LwF8OJps9r

— Le Monde (@lemondefr) June 16, 2021

[InsideLab] Les robots publient aussi...@gcabanac et Cyril Labbé se sont donnés pour mission de les détecter.

+ @Univ_Lorraine rejoint @DoRANum_
+ @ComsciconFra fait le buzz

Un test ? https://t.co/5h9YY78WLl (Vite plus que qq jours) pic.twitter.com/YQegupag35

— 𝖳𝗁𝖾𝖬𝖾𝗍𝖺𝖭𝖾𝗐𝗌 (@TheMeta_News) June 11, 2021

AUDIO ▶ Comment repérer les fausses études scientifiques ?

C'est dans le Billet sciences, à réécouter ici 🔊👇https://t.co/ACBzchvBek pic.twitter.com/iVKbY0gQ44

— franceinfo (@franceinfo) June 18, 2021

[#Podcast] Et pour (ré)écouter le #JournalDesSciences du jour > #Bételgeuse : lumière sur le mystère de son assombrissement https://t.co/8JudA6WZff par @NatachaTriou via @franceculture #LaMethSci pic.twitter.com/tAbnFajq5p

— Méthode Scientifique (@lamethodeFC) June 17, 2021

“[T]hese shit comments”: Author of a nonsense paper responds on PubPeer https://t.co/j2YYdwIF0U

— Retraction Watch (@RetractionWatch) September 22, 2021

Französische Forscher entdecken 243 Nonsens-Papiere https://t.co/C3ho9YZVvv

— FAZ Stellenmarkt (@FAZStellenmarkt) November 2, 2021

ENQUÊTE. Les études scientifiques, elles aussi victimes des fraudeurshttps://t.co/Th6Pk3loru pic.twitter.com/yJfGaIW1j6

— franceinfo (@franceinfo) November 6, 2021

Just out: “Prevalence of nonsensical algorithmically generated papers in the scientific literature” w/ Cyril Labbé in @JASIST https://t.co/N4wVqI8JHF. We report cases of SCIgen-generated publications. 😵 Companion website https://t.co/fSmzEq49x9 + thread: https://t.co/he2nEkzTpC pic.twitter.com/67x3XIuvFc

— Guillaume Cabanac (@gcabanac) May 26, 2021

Nonsensical research papers generated by SCIgen software are *still* popping up in the scientific literature many years after the problem was first seen. Could lead to 200+ retractions. Me on a study from @gcabanac & Cyril Labbé https://t.co/IWRRkU9bZv

— Richard Van Noorden (@Richvn) May 27, 2021

There's a long thread of misconduct cases, mostly in computer science, (https://t.co/ZdY51lvs8u) that we posted on @PubPeer (https://t.co/wboufpak0f). No reaction from the publishers involved to date AFAIK. Care for sunlight as disinfectant, anyone?

— Guillaume Cabanac (@gcabanac) April 16, 2021

This 2013 @IEEEorg conference paper has a typo in its title: 𝘢𝘭𝘰𝘨𝘳𝘪𝘵𝘩𝘮. Despite 56 ‘full text views’ according to the publisher, this paper was cited 252 times according to @Scopus. ⚠️Many citations come from #SCIgen-generated papers, such as https://t.co/O0bhe0RB9c pic.twitter.com/bjaotROAP0

— Guillaume Cabanac (@gcabanac) March 4, 2021

An @IEEEorg paper containing text generated by the ‘Automatic SBIR Proposal Generator’ (https://t.co/XGMZtSsKAT not affiliated to @NSFSBIR AFAIK, wink-wink) has been unnoticed since 2010 — and is still for sale. Details and links on @PubPeer https://t.co/I0zRfkPI9K pic.twitter.com/VlVLhYQkAo

— Guillaume Cabanac (@gcabanac) March 8, 2021

A nonsensical #SCIgen-generated paper for $107.19 at @ingentaconnect. It has one citation from another SCIgen paper now removed from @IEEEXplore without any retraction notice. Posted on @PubPeer https://t.co/lnDEhrsPhO pic.twitter.com/NJD1jaOLXT

— Guillaume Cabanac (@gcabanac) March 10, 2021

A paper containing #mathgen computer-generated text appeared in 𝘊𝘭𝘶𝘴𝘵𝘦𝘳 𝘊𝘰𝘮𝘱𝘶𝘵𝘪𝘯𝘨, a @SpringerNature journal. Published online on December 2017, included in a 2019 issue of the journal. Details & links posted on @PubPeer https://t.co/kl7rfLMtxq CC @SpringerEthics pic.twitter.com/JTwnBpIor8

— Guillaume Cabanac (@gcabanac) March 3, 2021

A #SCIgen computer-generated paper appeared in 𝘗𝘳𝘰𝘤𝘦𝘥𝘪𝘢 𝘊𝘰𝘮𝘱𝘶𝘵𝘦𝘳 𝘚𝘤𝘪𝘦𝘯𝘤𝘦 (@ElsevierConnect) in February 2021. https://t.co/B0msM4szCT pic.twitter.com/y0d6zyWJRM

— Guillaume Cabanac (@gcabanac) March 2, 2021

A @IEEEorg conference held in April 2021 published a #SCIgen-generated paper. The reference list does not appear to be randomly generated, though. Gibberish sold $33.00 or $14.95 for @ComputerSociety members. Details on @PubPeer https://t.co/Xrhk1GCBHD pic.twitter.com/BFffQuL7bH

— Guillaume Cabanac (@gcabanac) July 13, 2021

There are several entirely plagiarised conference papers by this author, but this is the first one with a paraphrased warning message from plagiarism detection software in the abstract. @IOPPublishing pic.twitter.com/2zT5GPY6UT

— Nick Wise (@nickwizzo) August 23, 2021

Editorial assessment of the 𝘈𝘳𝘢𝘣𝘪𝘢𝘯 𝘑𝘰𝘶𝘳𝘯𝘢𝘭 𝘰𝘧 𝘎𝘦𝘰𝘴𝘤𝘪𝘦𝘯𝘤𝘦𝘴: publishers should monitor time-to-market at the journal level and *react* to improbable shrinking durations popping up 😨. See https://t.co/EAEx5u4IoY for context and https://t.co/2oIsBw1nOD pic.twitter.com/nm2jF9XIyj

— Guillaume Cabanac (@gcabanac) September 29, 2021