Press Gazette Puts an AI Checkpoint Before the Byline

A new prepublication test for synthetic media

A newsroom safeguard moves upstream

For years, publishers have treated artificial intelligence as both a production tool and a post-publication problem. On September 10, 2026, Press Gazette and Dublin-based technology company Flip-Pay announced a different response: a system designed to inspect stories and media before they acquire a byline.

Called Made by Humans, the platform is being tested inside Press Gazette’s own content-management system. It checks text, images, video and audio for signs of artificial intelligence generation or manipulation, then gives editors a result such as “AI Generated,” “AI Assisted” or “Human Made.” The final decision remains with the editor.

The launch marks a significant shift in AI journalism. Instead of relying only on corrections, reader complaints or investigations after publication, publishers are beginning to place an authenticity checkpoint inside the ordinary editorial workflow.

Source: InPublishing — www.inpublishing.co.uk/articles/press-gazette-and-flip-pay-l...; Press Gazette — pressgazette.substack.com/p/new-tool-in-fight-against-ai-fak...

The problem is no longer theoretical

Press Gazette’s product follows a series of investigations into fabricated experts, invented identities and misleading material that appeared in established media outlets. The publication said its investigations identified more than 1,000 examples of fake or AI-generated experts entering mainstream news environments.

That record has sharpened the commercial stakes. A fabricated quote or synthetic image can trigger a correction, but the damage often arrives earlier: in the form of lost reader confidence, embarrassed advertisers and additional legal exposure for the publisher.

In a survey of more than 200 publishers, editors and journalists cited by Press Gazette, 95.3% said AI-generated material should be identified for readers. Yet 63.5% said their organizations lacked a reliable way to check for unauthorized AI use. The gap between what editors say they want and what their systems can actually verify is precisely the market Made by Humans is targeting.

For AI journalism, the implication is practical rather than philosophical. Detection is becoming part of publishing infrastructure, much like spell-checking, plagiarism screening and image-rights review.

Source: InPublishing — www.inpublishing.co.uk/articles/press-gazette-and-flip-pay-l...; Flip-Pay — www.linkedin.com/posts/flip-pay-emea_fomt26-futureofmediatec...

Detection helps, but it does not replace reporting

Made by Humans brings together specialized detection services, including Pangram for text and IdentifAI for visual and audio material. According to the companies, the platform can operate within a publisher’s CMS and create an audit trail showing what was checked before publication.

That design matters because AI detectors are not verdict machines. They produce signals, not proof. A human-written article can contain formulaic language that triggers a false positive, while a sophisticated synthetic story may evade detection. Editors still have to verify names, documents, images, sources and quotes through conventional reporting.

The most useful role for the software may therefore be triage. An editor handling a fast-moving story could receive an immediate warning that a photograph has been manipulated or that a submitted article deserves additional scrutiny. The system can slow down risky material without automatically blocking legitimate work.

This distinction is central to responsible AI journalism. The technology is being used to reinforce editorial judgment, not to outsource it. Press Gazette editor-in-chief Dominic Ponsford described the product as a way to prevent misleading material from being published, while Flip-Pay chief executive Paul McCarthy-Brain said it was intended to provide an evidence-based assessment and audit trail.

Source: InPublishing — www.inpublishing.co.uk/articles/press-gazette-and-flip-pay-l...; AI For Newsrooms — aifornewsroom.in/media

Why the timing matters for publishers

The launch comes as news organizations are experimenting with AI for transcription, translation, search, summarization and audience products. Those uses can reduce routine workloads, but they also expand the number of points at which synthetic material can enter the publishing chain.

A newsroom may use one system to summarize a public meeting, another to generate a headline and a third to create an audio version of the story. At the same time, freelancers, public-relations teams and outside contributors may submit material created with tools that are not disclosed. The editorial challenge is no longer simply whether a publisher uses AI. It is whether the publisher can account for where AI entered the process.

That is why provenance and labeling are becoming competitive issues as well as ethical ones. A visible “Human Made” badge could help publishers distinguish verified reporting from anonymous synthetic content, although its value will depend on transparent standards and consistent enforcement.

For smaller outlets, cost and integration will be decisive. A tool that sits inside an existing CMS may be more useful than a separate verification website, but detection services must remain affordable and accurate across languages, formats and newsroom styles. The risk is that wealthy publishers gain new layers of verification while local organizations remain exposed.

Source: InPublishing — www.inpublishing.co.uk/articles/press-gazette-and-flip-pay-l...; Press Gazette — pressgazette.substack.com/p/new-tool-in-fight-against-ai-fak...

The editorial test ahead

Press Gazette’s experiment will not settle the larger debate over artificial intelligence and news. It does, however, identify a more immediate question for publishers: can they prove that the material carrying their name has been checked before it reaches the public?

The answer will require more than a badge or a detector. It will depend on clear disclosure rules, trained editors, source verification and records showing how a story was assembled. In that sense, the most important feature of Made by Humans may be its audit trail rather than its label.

The next phase of AI journalism will be judged less by how quickly machines can produce copy than by how carefully newsrooms can control the boundary between assistance and fabrication. By moving verification before publication, Press Gazette and Flip-Pay are betting that trust is best protected at the point where a story becomes public.

That is a modest technological intervention, but a consequential editorial one: in an era of synthetic media, the byline may need a checkpoint of its own.

Source: InPublishing — www.inpublishing.co.uk/articles/press-gazette-and-flip-pay-l...; Nieman Journalism Lab — www.niemanlab.org/reading/press-gazette-launches-a-newsroom-...