A group of French research students is launching an online register to flag up scientific papers that have been tainted by fraud and other types of scientific misconduct.

Claire Ribrault, a PhD student in neurobiology at Ecole Normale Supérieure in Paris, unveiled the Scientific Red Cards project last month at a workshop on research integrity sponsored by the European Science Foundation (ESF). The idea is to identify papers that have been shown to be fraudulent but are still in circulation.

Scientific journals are the primary means by which the results of research are made public. Emma Campbell, of the European Association of Science Editors, told the meeting that editors were becoming more aware of their role as gatekeepers. Most journals now have guides to best practice for authors and reviewers and many require authors to sign a declaration before a paper is accepted.

The most important line of defense is peer review, where the submitted paper is subject to critical reading by a scientist familiar with the field. But by tradition reviewers are unpaid and may not be able to devote sufficient time and effort to detect signs of misconduct. Once a paper has passed peer review it can be difficult for editors to spot problems by themselves.

"How do you know if someone's fabricated their results if the reviewer didn't detect it?" Campbell asked. "How do you know if someone hasn't declared their conflict of interest?"

Modern tools available to editors can help them detect plagiarism in submitted articles and even whether digital images have been manipulated, but such services come at a price and evidence of misconduct may not be found until after the paper is published.

Guidelines to help editors and publishers deal with suspected misconduct are published by the Committee on Publication Ethics (COPE), as well as several other organisations.

"We retract a published paper if something has been found where perhaps the results have been fabricated," Campbell said. "Journals can ban submissions from someone in future if they decide that person is not someone they can trust to submit decent papers. And they can contact the author's employer or other authorities, for example the British Medical Association."

On the other hand, there is no formal system to prevent a paper discovered to be fraudulent from being resubmitted to a different journal, she said. She called for a consensus amongst publishers on the guidelines to be used. "Rather than having lots of different sets of guidelines in different professional groups, try and have one set of international guidelines for people to follow."

But once a fraudulent paper has been published it is very difficult to remove it. Journals can retract articles from their online databases but libraries all over the world are stocked with printed journals that cannot be recalled.

It was how to tackle this problem of fraudulent papers remaining in circulation that prompted Ribrault and her colleagues, all PhD students in life sciences, to set up their Scientific Red Cards website.

As a research student she was concerned that there was a lot of informal comment on which papers could not be trusted but little reliable information. "There were lots of people saying don't trust this paper because it has been falsified, don't trust this guy because he has been hiding a conflict of interest," she told the meeting.

She pointed out that even where a published paper is found to be tainted by fraud "sometimes the paper is not retracted, depending on the policy of the journal, and even if the paper is retracted sometimes it's still cited after the retraction."

The database will contain the bibliographic reference of the paper, the type of misconduct, and a link to a published account of the misconduct. Papers affected by falsification, fabrication and plagiarism will be included as well as cases where editorial policies and standards have not been respected or research subjects have been treated unethically. Ribrault said they had listed 30 papers so far.

"We think better visibility of misconduct for scientists could promote scientific integrity," she said. "Making scientific misconduct known to the scientific community could act as a deterrent to other misbehaviour."

The meeting gave a cautious welcome to the proposals but pointed out that unless the students were careful they could expose themselves to legal problems. There was also concern that the reputations of innocent co-authors might be tarnished. Others said that the project was in the tradition of the self-policing of the scientific community and should be supported.

Ribrault stressed that access to reliable scientific information was essential for well-informed democratic decision-making. "We are all interested in the relation between science and society and the impact science has on society."

The meeting in Madrid on 17-18 November was organised by the newly formed Research Integrity Forum of the European Science Foundation (ESF) in collaboration with the Spanish National Research Council (CSIC). It continued work set in motion by the first world conference on research integrity held in Lisbon in September 2007.