AI in journalism: A running record of scandals and mistakes
News outlets worldwide have published AI-generated content, fabricated quotes, and fake authors over the past three years. The incidents range from careless editing oversights to coordinated deception by people submitting AI work under false identities.
These cases matter for writers because they show where editorial systems fail-and what happens when they do. They also demonstrate how quickly AI tools can fool experienced journalists and editors.
Recent cases (2026)
Telegraph leaves AI editing note in published article (May 2026). A story about Donald Trump and Xi Jinping contained an embedded instruction from an AI chatbot: "To further divide the piece and maintain that authoritative, broadsheet pace, here are two additional subheads." The paragraph was removed after publication. The Telegraph declined to comment.
Multiple outlets publish fake Thai police image. The Telegraph, Sun, Mirror, Express, GB News, Daily Mail, New York Post, and Daily Star all ran the same story about Thai police officers who supposedly went undercover in drag to catch a drug dealer. The image was AI-generated. Thailand's Tha Luang police station later posted: "The real one is here, everyone. It's AI."
New York Times publishes AI-generated quote as real. The paper attributed words to Canadian opposition leader Pierre Poilievre that were actually an AI-generated summary of his views. The correction came more than two weeks after publication. The supposed quote included the phrase "turncoats"-a word Poilievre had never used. ChatGPT Courses teach journalists how to verify AI outputs, a step this reporter skipped.
Mississippi Free Press discovers fake author. An opinion column about the gig economy was written by someone using a fabricated identity. The editor grew suspicious only when the author submitted an invoice with a different name. He then discovered the headshot was AI-generated, social media links were dead, and other columns from similar "new authors" showed matching red flags. The outlet is developing a formal AI policy and staff training.
New York Times ends relationship with reviewer over AI plagiarism. A freelance book reviewer used ChatGPT to help write a review, and the tool incorporated language from a Guardian review without attribution. The reviewer admitted the error and apologized to both the Times and the Guardian.
Ars Technica fires AI reporter for fabricated quotes. A senior AI reporter at Conde Nast's tech site published quotes attributed to a real person who had never said them. The reporter used ChatGPT to extract source material but failed to verify the results against the original. He said he was working while unwell. The outlet has updated its AI guidelines to prohibit using AI tools to generate, extract, or summarize material attributed to named sources.
Mediahuis Ireland CEO publishes fabricated AI quotes. Peter Vandermeersch, former chief executive of Mediahuis Ireland, admitted using ChatGPT, Perplexity, and Google Gemini to summarize reports, then publishing quotes from those summaries as real. An investigation found unverifiable quotes in 15 of 53 blog posts. Vandermeersch said he fell into the trap of trusting AI summaries without checking the original sources.
Crikey removes articles over undisclosed AI use. Australian news site Crikey took down four articles after a contributor used ChatGPT for proofreading, sense-checking, and rephrasing-against the outlet's editorial policy. The contributor was not sent the editorial guidelines before submitting. Crikey said it needs clearer communication with new contributors about AI restrictions.
2025 cases
Victoria Goldiee articles removed across four publications. The Local Toronto investigated a freelance writer who had placed articles in The Guardian, Outrider, Dwell, and the Journal of the Law Society of Scotland. The editor who commissioned her became suspicious about her location and sources. When he contacted people quoted in her previous work, they denied speaking to her. Her pitch and follow-up emails showed "rote phrasing" typical of AI output. The Guardian removed a first-person essay about music sharing; Outrider removed a climate change article; Dwell retracted a home design piece; the Law Society removed an article with disputed quotes.
Business Insider removes 38 essays. The outlet removed 38 first-person essays after investigating concerns about author identity and veracity. The investigation began after Press Gazette reporting on another fake freelancer. Editor-in-chief Jamie Heller said the outlet had bolstered verification protocols.
Wired and Business Insider pull 'Margaux Blanchard' articles. Several publications removed articles by freelancer "Margaux Blanchard" after Press Gazette found her stories contained unverifiable case studies and named people who did not appear to exist. Wired said the story "did not go through a proper fact-check process or get a top edit from a more senior editor." The publication later confirmed it was an AI fabrication.
Elle and Marie Claire Belgium use fake AI journalist profiles. Ventures Media, which publishes these titles in Belgium, created fake journalist profiles with AI-generated photos. Articles by profiles named 'Claire De Wilde' made up nearly half of Marie Claire Belgique's 2025 output. After exposure, the publisher changed the bylines to outlet names and added disclaimers stating content was AI-generated and reviewed by editors.
The Grind delays food issue over AI pitches. Toronto magazine The Grind greenlighted several article pitches it later identified as AI-generated. Editors became suspicious when a writer claimed to have interviewed subjects at immigrant-run restaurants but was based outside Canada. When challenged, the writer admitted the "characters and places" were "fictional composites." Seven of the commissioned articles showed signs of AI writing: U.S. spelling instead of Canadian, double-barreled headlines, vague details, and fake contact information for sources.
Chicago Sun-Times and Philadelphia Inquirer publish nonexistent books. A summer reading list produced by King Features and used by both outlets contained books that do not exist. A freelance journalist created the list using an AI agent without disclosing the use. The Sun-Times said the content "was inserted into our paper without review from our editorial team." Both outlets removed the section and updated policies. King Features terminated its relationship with the journalist.
Press Gazette reveals fake experts in media. Dozens of stories were removed or amended after reporting on fake experts being quoted across UK and US publications. Companies selling CBD oil, sex toys, and vapes were apparently using AI to create expert personas and game search rankings. Many fake experts used journalist response services to answer queries with AI-generated replies.
Harper's Bazaar Netherlands publishes AI-written book review. The Dutch edition published a top-five books article written with AI. An author whose book was reviewed noticed factual errors and confirmed the piece matched ChatGPT's output about his work. After the publisher was contacted, it confirmed AI was used and said this violated internal guidelines.
Earlier cases (2023-2024)
Sports Illustrated publishes AI-generated writers. The outlet was accused of publishing AI-written articles after several authors were found to have AI-generated headshots and fake names. One author photo was for sale on a website selling AI headshots. Sports Illustrated said the content came from a third-party company, AdVon Commerce, and removed the articles. The Sports Illustrated Union said staff were "horrified" by the reporting.
CNET removes AI-generated articles. The tech news site removed articles written by its own AI tool after errors were identified. Of 77 stories published using the tool, 41 needed corrections. Issues included incomplete company names, transposed numbers, and vague language. CNET paused the program and said it would restart only when confident in both the tool and editorial processes.
What writers should know
These cases share common patterns. Fake authors use journalist response services to answer pitches quickly. Real writers use AI tools without disclosing them. Editors fail to verify quotes against original sources. Outlets don't flag third-party content clearly. New contributors don't receive editorial guidelines before submitting.
The mistakes harm readers, sources, and the outlets themselves. They also damage trust in bylines-the basic signal that a real person wrote and stands behind the work.
For writers, the lesson is straightforward: disclose any AI use to your editor before submitting. Verify every quote against the original source. Check that people you quote actually said what you attribute to them. If you use AI to summarize, paraphrase, or extract information, treat those outputs as rough notes, not finished facts. Read AI for Writers resources to understand how to use these tools responsibly.
Outlets are tightening verification processes, updating policies, and training staff. But the responsibility starts with the writer.
Your membership also unlocks:
