A new study in The Lancet shows that the rate of fake citations increased more than 12x between Jan 2023 and Feb 2026.https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(26)00603-3/fulltext

[petersuber@fediscience.org]

A new study in The Lancet shows that the rate of fake citations increased more than 12x between Jan 2023 and Feb 2026.
https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(26)00603-3/fulltext

#AI #Citations #Hallucinations

[noplasticshower@infosec.exchange]

@petersuber this is going to be an aspect of recursive pollution. #MLsec

Recursive Pollution hits the CISO Circuit | BIML

Cynthia Brumfield wrote an excellent, in-depth article for CSO on recursive pollution that is well worth a read. At BIML

Berryville Institute of Machine Learning (berryvilleiml.com)

[hydreprever@mathstodon.xyz]

@petersuber ️ La recherche biomédicale est vraiment l'homme malade de la science...

[flomaraninchi@pouet.chapril.org]

@HydrePrever

A chaque fois qu'on me dit que l'IA c'est formidable parce que ça va soigner toutes les maladies... Oui, bon, en attendant ça bordélise la recherche biomédicale, surtout.

The purpose of a system is what it does.

@petersuber

[djelouze@mstdn.calut.fr]

@flomaraninchi @HydrePrever @petersuber

"Flagged references underwent sequential filters to minimise false positives: automated pattern detection removed parsing artefacts, and an LLM (Claude 3.5 Haiku; Anthropic, San Francisco, CA, USA) screened remaining candidates to distinguish genuine fabrications from formatting discrepancies such as informally abbreviated titles."

I don't know what level of trolling is this

[flomaraninchi@pouet.chapril.org]

@djelouze

@HydrePrever @petersuber

[hydreprever@mathstodon.xyz]

@djelouze @flomaraninchi @petersuber

[dckim@mastodon.social]

@petersuber I guess it might be time to go back to the in-house hard-copy to determine what is true or false. We're seeing the death of information and the dumping of concocted half-truth of the most misleading kinds. That coupled with human over-confidence and over-reaching is a bit troubling.

[nicelymanifest@mastodon.social]

@petersuber Threshold of trust long since breached ...

[jcblubaugh@mastodon.social]

@petersuber but can we trust this?

[knutson_brain@sfba.social]

@petersuber
Potential incentized correction : Discovery of a fake citation during review leads to instant rejection.

[xs4me2@mastodon.social]

@petersuber

So much for LLM and AI, but of course authors used to make up things long before…

[shironeko@fedi.tesaguri.club]

@petersuber fake citations in published work or submitted work

[kitkat_blue@mastodon.social]

@petersuber

Thanks Gen-ai! You are truly a gift to science and the world!

[shironeko@fedi.tesaguri.club]

@petersuber looks like it's published work. All of this just shows how fake and broken the publishing industry is.

[clickhere@mastodon.ie]

@petersuber zomg no way

[clickhere@mastodon.ie]

@djelouze @flomaraninchi @HydrePrever @petersuber

We are now, truly, in the Dark Ages.