The use of artificial intelligence in editorial practice brings with it increasing risks to media credibility. The British server Press Gazette has launched a special project that tracks cases of AI failures involving fictional authors, plagiarism and fabricated citations. According to the analysis, even renowned titles such as The New York Times or non-profit organizations like Mississippi Free Press have run into trouble.
The latest recorded case from April 2026 concerns the American media outlet Mississippi Free Press (MFP), which unknowingly published an article about the sharing economy written by artificial intelligence. The editorial office discovered the deception only when the author sent an invoice with a name that did not match the signature details. Subsequent checks revealed that the author's digital footprint was non-existent and his profile photo had been generated by an AI system.
“It was no longer about doubts about the text itself, but about the author. All social media links were non-functional. It turned out that the author tried the same trick on another editor,” said Kevin Edwards of MFP.
A serious breach of standards also occurred at The New York Times. The editorial office terminated its collaboration with external reviewer Alex Preston after readers pointed out a striking similarity between his review of the book “Watching Over Her” and a text in The Guardian. Preston admitted that he had used an AI tool to prepare the draft, which incorporated passages from another review into the text, and he failed to remove them before submission.
Even experienced journalists face problems. Peter Vandermeersch, former CEO of Mediahuis Ireland and former editor-in-chief of the Dutch newspaper NRC, was suspended after so-called artificial intelligence hallucinations appeared in his texts. Vandermeersch used tools such as ChatGPT and Perplexity to summarize extensive reports, which he then cited. However, analysis revealed that in 15 of 53 posts, citations were fictional or mistakenly attributed to other persons.
“I put words into people's mouths that they never said. It was more than careless, it was wrong,” Vandermeersch admitted on his blog.
Other cases include the Australian website Crikey, which had to delete a series of articles in March 2026 after Professor Toby Walsh pointed out the obvious use of AI without acknowledging the source. The editorial office confirmed that the author used ChatGPT for editing in a way that violated strict internal rules for using technologies. Press Gazette states that these incidents serve as a warning to publishers that relying on automatic AI detectors is often ineffective and requires increased human oversight.