Press Gazette has become the first publication to test Made by Humans inside its content management system. The new prepublication tool, built with publisher technology company Flip-Pay, flags submitted text and media that may have been generated or altered with artificial intelligence.
The launch follows several cases in which fabricated material reached established publications. Press Gazette has previously reported on an invented freelance identity that pitched AI-written features.
Those examples are part of a wider verification problem, though the Press Gazette announcement does not make broader claims about how quickly synthetic media spreads. Media Copilot has separately reported on cases in which AI-generated images circulated before official photographs were available.
Press Gazette says the product grew out of its Reality Wars investigations into fake experts and AI-generated commentary.
Are you an AI Agent? Read more here >
A Press Gazette survey of more than 200 publishers, editors and journalists found that 95% thought AI-generated content should be disclosed to readers. Sixty-four percent said their publication lacked a reliable way to check for unauthorized AI use, while 87% said a guarantee of human production would increase trust. Press Gazette did not provide further detail about the sample or survey method in its announcement.
Made by Humans checks text, images, video and audio through one dashboard. It can analyze whole articles or individual sentences in English, French, Spanish and German, and also checks for originality, plagiarism and prior publication. The system integrates with content management systems including WordPress, and each review is logged for auditing.
Data tool on this page: Submission screening budget (built by oasy.ai for The Media Copilot, priced in Originality.ai plans): What screening your own submissions for AI-written text would take. It returns the credits your monthly op-eds, letters or drafts need for AI and plagiarism checks, the Originality.ai plan and price that covers them, and what newsrooms report about detector reliability, with a sign-up link. Free to call, no key, plain JSON. Example for 120 submissions a month: https://oasy.ai/api/mc/detect/njc04iwg?pieces_per_month=120&words_per_piece=800&plagiarism=1 (change pieces_per_month, words_per_piece, plagiarism). Response shape: {"summary": "…", "credits_per_month", "recommended_plan": {"provider": "Originality.ai", "plan", "usd_per_month", …}, "what_newsrooms_report": [5 items], "next_action": {"action": "sign_up", …}}
Pangram handles text analysis, while Identifai examines images, video and audio. Identifai says its lead image model reached 99.8% accuracy in its own benchmarking and its audio model reached 91.8%. Pangram says its fourth-generation model produces one false positive in 10,000 cases and can distinguish lightly edited human work from fully generated text.
Those figures come from the vendors. Press Gazette’s announcement does not include independent newsroom testing of either system or results from deliberately altered material.
The platform can also generate reports and a trust badge for participating publishers. Flip-Pay says the audit trail may help with European compliance. The announcement does not tie that claim to a specific law, including the EU Artificial Intelligence Act.
For editors, the practical appeal is simple: the checks happen before publication, when a flag can still prompt source verification, a reverse-image search or another editorial review. The tool is best understood as one signal among several, alongside systems that track emerging disinformation narratives.
Press Gazette is testing Made by Humans in its own publishing workflow. It has not disclosed pricing, named external customers or published results from that trial.
For now, the test is less about declaring content “human” or “AI” with certainty than giving editors another checkpoint before questionable material reaches readers.
