No menu items!

Trending: criminal scientists cheat with artificial intelligence

By Martina Frei

“These schemes are spreading like cancer. We are heading for a crisis. We can’t just let this continue,” says Bernhard Sabel, and in doing so, the Institute of Medical Psychology director at Otto von Guericke University Magdeburg, Germany, sounds seriously concerned.

At the end of 2020, the professor heard about “paper mills” for the first time. These writing rooms, of which no one knows who is behind them, offer their services to scientists.

The customers can choose those who have completed their research project and hand over their data to the “paper mill,” which then writes the manuscript and arranges for publication in a scientific journal.

“That costs about 1,000 euros,” says Bernhard Sabel, who has looked at various offers.

For 26,000 euros, you can get a freely invented “scientific” publication

For around 8,000 euros, the “paper mill” unceremoniously creates a manuscript, writes it, and publishes it in a scientific publishing house.

The customers act as authors.

“All the prospective author has to do is name a specific field, possibly include a few keywords or methods, and select a journal,” according to an article in “Labor journal.”

According to Sabel, the “all-around package” is available for 17,000 to 26,000 euros (US$17,000 to 26,000).

At the end of 2020, the professor heard about "paper mills" for the first time. These writing rooms, of which no one knows who is behind them, offer their services to scientists. (Photo internet reproduction)
At the end of 2020, the professor heard about “paper mills” for the first time. These writing rooms, of which no one knows who is behind them, offer their services to scientists. (Photo internet reproduction)

For this price, a “paper mill” provides the design for a research project, supposedly conducts the experiments – which in reality never take place – writes a manuscript with the invented data, inserts pictures and graphics, and sends it (contrary to general practice) to several scientific journals at the same time – and gets the go-ahead for publication from an editorial office.

With more than 50,000 scientific journals, the choice is vast.


“The more prestigious the journal, the higher the price,” says Sabel.

“To be sure, fakes have always existed and always will. But the mass, global, industrial production of completely fabricated scientific articles – that’s new and very worrying. In recent years, an entire industry has developed there.”

These fake studies and articles are written by artificial intelligence (AI) trained on millions of articles. Sometimes scientists provide editorial assistance.

“The texts are so sophisticated that no one can tell anymore.”

Bernhard Sabel, director of the Institute of Medical Psychology at Otto von Guericke University in Magdeburg, Germany

“I was shocked to learn at a recent congress how well AI writes such technical articles,” Sabel says. “In the past, manuscripts written by AI still contained linguistic or logical errors – now the texts are so polished and of such high quality that no one can tell anymore.”

Another ploy of the “paper mills”: they translate Russian technical articles, for example, and submit the translation to an English-language journal.

Sabel knows of an AI test in the U.S. in which a scientific publication that helped Italian nuclear physicist Enrico Fermi win the Nobel Prize in 1938 was translated with AI, edited, and sent to a prestigious journal.

“It was accepted as worthy of publication, but not published because the whole thing was only meant as a test.”


Paper mill articles were a big problem, especially in medicine and computer science.

“These are not isolated cases,” says Sabel, who is involved with the issue on the extended executive committee of the German Academic Association.

]He says dozens of other disciplines are also affected, including psychology, sociology, business administration/marketing, agricultural sciences, and philosophy.

Shortly after he learned about “paper mills,” Sabel discovered that 10 to 15 of about 200 articles reviewed had been problematic in the neuroscience journal he is editor-in-chief.

“We were more affected than I could have imagined. It did worry me.”

Sabel estimates that about ten percent of published articles in neuroscience journals are “highly suspect.”

Clear proof that a paper comes from a “paper mill” is only possible in individual cases. In most cases, this is not known with certainty, says Sabel.


He took random samples with his colleagues. Of 1,000 articles published in ten medical journals, 238 were suspected to be fabricated. “These papers were mainly from China, India, Iran, Turkey, and Russia.”

According to Edwin Charles Constable, president of the Swiss Academies of Arts and Sciences Expert Group on Scientific Integrity, entirely fabricated papers would likely be uncovered through standard peer-review procedures or post-publication peer evaluation.

“When you read a paper like this, you often feel something is wrong.”

heading for a crisis, Trending: criminal scientists cheat with artificial intelligence

Meanwhile, the psychologist and his team have scrutinized more than 13,000 scientific articles.

“The results surprised us,” he says, but he doesn’t want to reveal more until the work has been reviewed by independent reviewers and published.

Only this much: “The problem is fundamental, and it’s been growing for ten, fifteen years.”


So-called “open access publishing” – free online access to scientific articles without expensive journal subscriptions – also contributes to this.

Thanks to free access, much more can be published with little effort, without editorial, distribution, and printing costs.

“Everyone was happy about the increase – including scientific publishers, academic institutions, and states,” says Sabel.

That “open access” was also wind on the “paper mills” was something no one expected.

“Paper mill detectives estimate that thousands or tens of thousands of alleged scientific publications are merely fictitious – yet all of these fantasy articles passed the peer-review process by reviewers from internationally accepted scientific journals.

Given the 2.8 million scientific articles published yearly – and rising – even a few tens of thousands of fakes don’t seem like much.

But no one knows the number of unreported cases. Some estimates even put the number of fabricated articles at over 400,000 per year, about half of them in biomedicine.


And if such a “paper mill” article is not accepted for publication somewhere, it is published in another journal.

This resulted from the follow-up of 13 dubious articles previously rejected by the publisher “FEBS Press”.

According to an employee of “FEBS Press,” referred to by the science magazine “Nature,” the number of papers originating from paper mills has increased massively in recent years.

In the case of the scientific journal “Molecular Brain,” every fourth article submitted for publication was presumably fictitious.

The scientific publisher Elsevier told “Nature” that its employees identify thousands of “paper mill” articles yearly and fish them out before publication.


“Paper mills” customers come from China, Russia, Iran, Japan, India, Korea, the United States, and other countries.

“Chinese scientists working in laboratories here on Chinese government grants must sign contracts in China. They must pay back the grant if they don’t meet certain goals. Such targets are often publications.”

“Parents or relatives stand as ‘guarantors’ on these contracts. One can imagine what happens if the publications fail to materialize during such a research stay.”

“This can lead to ‘fabrication’, sometimes obvious with the acquiescence of the laboratory director, who wants to help the employee save face,” explains Ulrich Dirnagl, founding director of the “Centre for Responsible Research QUEST” in Berlin.

Dirnagl has been working there for years on how to increase scientific quality.

With more than 50,000 scientific journals, the choice to publish fake papers is vast. (Photo internet reproduction)
With more than 50,000 scientific journals, the choice to publish fake papers is vast. (Photo internet reproduction)


The “paper mills” probably began to grind in about 2010. “China wanted to become the leading scientific nation,” says Sabel.

Accordingly, he says, the pressure on Chinese academics increased enormously. “But the whole thing is also a problem in other countries, but not to this extent.” Chinese physicians are a particular target group of the “paper mills.”

According to the science magazine Nature, the Beijing health authority decreed in 2020 that only the first authors of at least three scientific articles can become chief physicians.

At least two such first authorships are needed for promotion to the deputy, but there is no time in the daily hospital routine to earn academic merit.

Often, ten or more authors from the same institution would share the cost of the “paper mill,” Sabel knows.

Sometimes the alleged authors also come from departments that have nothing to do with the topic of the publication.

In Switzerland, fabricated publications “have not been a problem at all so far, in my opinion. But they could become one in the future,” Constable says.


The fundamental evil, experts agree, is the pressure on researchers worldwide. To make a career, you must publish heavily in the highest-ranking journals possible and be cited by others.

Authors would often respond to the real or imagined publication pressure they face from their institutions.

Optimally, Constable says there should be a coordinated response at the international academies and funding agencies level, “but as always, the devil is in the details.”

“We should bring publication pressure back down to a normal level and evaluate not only quantity but above all quality,” Sabel warns.

“The consequences of criminally acquired publications are disastrous because they undermine the credibility of science.”

“These damages trust in science. It also damages the reputation of academic institutions. And it ‘pollutes’ our world knowledge. The danger is that we won’t be sure which scientific results are true or made up later.”


In addition, the whole thing is spreading further and further when unsuspecting scientists cite such fake publications in their work, refer to them, and possibly start new studies with cancer, Alzheimer’s, or other patients based on them.

A single case could touch tens, hundreds, or thousands of articles at various publishers, “Nature” quoted a spokesman for the “PLOS” publisher.

Fabricated publications could also affect economic development, Sabel fears.

If, for example, African countries no longer knew which fertilizer was best for a particular soil condition because fake papers presented false results, this could directly impact harvests.

Thanks to committed editors-in-chief like Sabel, “the full extent of this waste” is now coming to light, Dirnagl wrote in the “Labor Journal”.

But experts see it differently: what is now becoming visible is almost certainly just the tip of the iceberg.


Sabel, for example, assumes that perhaps one percent or less of the faked papers are discovered and retracted.

He said that if intelligent “paper mill” detectives expressed suspicion, the chances of retraction were higher.

About one in four suspicious publications was retracted, according to research by “Nature.” That led to hundreds of retractions in the last two years.

Compared with previous years, retractions thus reached record numbers.

The European Review for Medical and Pharmacological Sciences has retracted over 180 articles since 2020.

The world’s venerable and premier pharmacology journal, Naunyn-Schmiedeberg’s Archives of Pharmacology, retracted more than 300 articles because authors did not provide original data when requested.

According to the report, 30 percent of all articles submitted to this journal in 2020 were fabricated.

Some withdrawn articles were later published in other journals with slightly different titles and authors.

Accordingly, he says, the pressure on Chinese academics increased enormously. (Photo internet reproduction)
Accordingly, he says, the pressure on Chinese academics increased enormously. (Photo internet reproduction)

In January 2021, “The Royal Society of Chemistry Advances,” a journal of the “Royal Society of Chemistry,” withdrew 68 supposedly peer-reviewed articles. All of the papers were by authors from China.

In October, the “Journal of Cellular Biochemistry” retracted 129 articles originating from “paper mills.”

In December 2021, a mass retraction occurred at “SAGE”: 122 papers were affected. In February 2022, “IOP Publishing” retracted 350 documents in one fell swoop, and in September 2022, the number rose to nearly 500, “Retraction Watch” reported.


Unfortunately, however, many journals still don’t even care about the problem or delay the necessary retractions, Dirnagl criticized.

“The publishers are not always particularly helpful,” confirms Constable. And the authors of these publications are usually uncontactable.”

Sabel’s experience shows that most scientists have never heard of this problem. He says even the president of a major U.S. medical society with 50,000 members reacted with amazement.

When Sabel explained the whole thing to him, he said: “Wow, I had no idea. In German: “I had no idea about that.”

This post was published first here.

Check out our other content