A peer reviewer rejected a paper by insisting that GPT-5 did not exist
In November 2025 a reviewer at a Springer Nature journal dismissed a submission on the grounds that GPT-5 was imaginary, calling the study's method impossible. The author appealed, wondering whether a machine had written the review. The journal found no sign of AI and called the GPT-5 slip a very minor element.
The publisher behind that journal is one of the giants of academic publishing, though its current form is young. Springer Nature was created in May 2015 by merging Springer Science+Business Media with Holtzbrinck's Nature Publishing Group, Palgrave Macmillan and Macmillan Education, with Holtzbrinck holding a 53 percent majority. Its roots are far older: Julius Springer founded Springer-Verlag in Berlin in 1842, Macmillan dates from 1843, and the journal Nature has been published since 1869.
The combined company has tried twice to float on the stock market, in May 2018 and autumn 2020, and abandoned both attempts because markets were unfavourable. It has kept expanding, buying the Paris open-access publisher Atlantis Press in 2021, and has spun off a stream of themed titles such as Nature Climate Change, Nature Energy, Nature Food and Nature Water. It says it became carbon neutral in 2020 and groups its output into 17 hubs tied to the UN Sustainable Development Goals.
Quality control is a recurring headache for any publisher of this size. In 2019 it retracted a paper over a dubious review process, and in November 2021 a lapse in reviewing at one geoscience journal forced it to withdraw 44 papers that made no sense at all. In August 2023, following an investigation, it withdrew a paper that claimed there was no evidence of a global climate crisis. In 2026 it began attaching expressions of concern to books while integrity investigations are under way.
Artificial intelligence adds new wrinkles. Reviewers are told to keep submitted manuscripts out of chatbots, both to protect sensitive material and because such software has limits, though in April 2026 it told the Committee on Publication Ethics that using AI to polish a review's language might be acceptable case by case.
Source: Springer Nature
