Press Gazette launches AI content checker for editors

Press Gazette and Flip-Pay have launched Made by Humans, a newsroom tool that checks whether content is AI-generated or manipulated before publication. The platform is aimed at editors working under new transparency rules in the European Union.

From within a publisher's content management system, it checks text, images, video and audio, classifying material as AI Generated, AI Assisted or Human Made. Editors can then decide whether to publish, review or label the content.

The launch comes as publishers face growing pressure to identify synthetic material in editorial workflows. Article 50 of the EU AI Act requires editors to label AI-generated content in public-interest journalism if it has not been closely reviewed by a human, with fines for non-compliance.

The product grew out of Press Gazette's Reality Wars investigations, which identified more than 1,000 published articles in mainstream media that cited fake or AI-generated experts. A survey of more than 200 publishers, editors and journalists found that 95.3% believed AI-generated content should be flagged to readers, while 63.5% said their publication had no reliable way to check for unauthorised AI use.

Made by Humans was built by Dublin-based Flip-Pay from a concept developed by Press Gazette Editor-in-Chief Dominic Ponsford. The platform has been trialled within Press Gazette's own content management system.

How it works

The product uses third-party detection tools, including Pangram for text analysis and identifAI for image, video and audio checks. It also offers plagiarism screening, reverse-image checks, audit trails and reporting functions intended to support internal compliance processes.

According to the companies, the service works in English, French, Spanish and German. It also integrates with WordPress and other content management systems for pre-publication checks.

Ponsford said the tool was designed as a safeguard for editorial teams, not a substitute for newsroom judgment.

"We named our investigations into AI-generated content 'Reality Wars' because quality, human-authored journalism is in a fight for survival against misleading AI-written bilge.

"We've seen hundreds of stories taken down by reputable titles after elements were identified as misleading or AI-generated. This tool will help stop such articles before publication.

"Made by Humans helps protect publishers from the colossal reputational damage that could come from publishing misleading AI-generated material, and is a tool to support, not replace, human editorial judgment.

"Like many publishers, we use AI for things like transcriptions, research and data analysis. But we pride ourselves on ensuring that human creativity, originality and curiosity underpin our journalism. This technology helps verify that all our content is 'human-made'," said Dominic Ponsford, Editor-in-Chief, Press Gazette.

Growing concern

The companies behind the service argue that synthetic media is becoming harder to identify through manual review alone. identifAI said it has logged around 10,000 deepfake incidents reported in English-language media or on social platforms since January 2020, with monthly reported incidents rising from four to more than 3,000 over the period tracked.

The same dataset found that 47% of those reported incidents were weighted towards the United States, while 51% appeared on X. Around 24.6% were geared towards political manipulation, according to identifAI.

Paul McCarthy-Brain, Chief Executive Officer of Flip-Pay, said publishers were dealing with both editorial and commercial risks as synthetic content spreads.

"We work closely with publishers around the world and can see how AI is undermining their commercial authority and, now, the quality and reputation of their editorial content too.

"Made by Humans uses best-of-breed AI-detection technology and covers every content risk through one platform, which can be embedded directly into the workflow.

"It offers editors a clear, evidence-based assessment with an audit trail that can help support EU compliance. It's a shining example of how AI can be used to benefit publishers," said Paul McCarthy-Brain, Chief Executive Officer, Flip-Pay.

Pangram said its text model can identify content produced by systems from OpenAI, Anthropic, Google, Meta and DeepSeek, and can distinguish between fully generated text and human writing that has been lightly edited with AI. identifAI said its lead image model detects AI-generated images with 99.8% accuracy in its own benchmarking, while its audio model reached 91.8%.

Kate Burns, Vice President, EMEA, at identifAI, said editorial teams were increasingly exposed to fabricated media.

"Almost anyone can now create a deepfake, and the most dangerous ones look exactly like the real thing. Many are picked up by hard-working fact-checkers and picture desks before they make it into the public domain, but editorial teams are on the front line here.

"Made by Humans, supported by identifAI, gives media and broadcasters the vital tools they need to ask the simple question - 'is this real?' - and greater peace of mind to publish," said Kate Burns, Vice President, EMEA, identifAI.

Max Spero, Chief Executive Officer of Pangram, said the rise of synthetic text had created a new transparency issue for publishers.

"AI-generated content is proliferating rapidly across many industries, and publishing is no different. We're proud that Pangram is integrated with Made by Humans to increase transparency for media organisations.

"Pangram 4 is our most sophisticated and accurate detection technology yet. It offers sentence-level AI detection for editors and is even more accurate. It also has built-in recognition against adversarial efforts such as humanisers.

"Importantly for editors, Pangram 4 also distinguishes between AI-polished human writing and full AI generation, allowing for journalists who might be using AI for tasks such as transcripts, research or light editing," said Max Spero, Chief Executive Officer, Pangram.