A handful of publishers have begun using artificial intelligence tools to spot instances of duplicated images in scientific manuscripts, Nature News reports.
For instance, it says that the American Association for Cancer Research checks papers it publishes in its journals using a service from Proofig, while other publishers are testing other tools or are developing their own, like the publisher Frontiers is. Others, though, are not convinced the benefits are yet worth the cost, it adds.
Image manipulation experts tell Nature News that it is currently difficult to compare tools aimed at spotting image manipulation, as there are no public, standardized comparisons of the different tools, and that drawbacks include that current tools mostly check for duplications within one manuscript rather than for a range of manipulation types and across publications. Still, the approach may eventually become widespread, as tools to check for plagiarism have, it adds.
Bernd Pulverer, chief editor of EMBO Reports, cautions that reliance on AI tools may then lead fraudsters to develop even better manipulation approaches. "I am concerned that we are entering an arms race with AI-based tech that can lead to deepfakes that will be impossible to find," he adds there.