An OpenAI-linked news outlet appears to be entirely AI-generated

Published: (April 26, 2026 at 12:05 PM EDT)
3 min read

Source: Mashable Tech

Overview

A new report from The Midas Project’s Model Republic publication finds that the news site The Wire by Acutus (acutuswire.com) relies almost entirely on AI‑generated content. The site has been operational since the end of 2025 and has published nearly 100 articles covering tech, energy, media, science, business, and healthcare.

Claims of “Collaborative Journalism”

The site’s About page describes its work as “collaborative journalism” led by an “editorial team,” yet it provides no masthead and credits no editors or journalists. The explanation for this anonymity appears in the “How It Works” subhead:

Our editorial team identifies timely topics and invites contributors with relevant, firsthand experience to share their perspective through structured conversations. Those perspectives are synthesized and edited into stories that reflect where contributors align, where they diverge, and what it all means — offering depth, balance, and clarity beyond the headline.

Findings from AI Detection

Journalist Tyler Johnston ran the site’s content through Pangram, an AI detection tool that claims 99.98 % accuracy. The results were:

  • 69 % of the 94 articles flagged as fully AI‑generated
  • 28 % flagged as partially AI‑generated
  • Only 3 articles classified as human‑authored

Content Bias

Johnston noted that the articles overwhelmingly favor the development of artificial intelligence and dismiss its critics. Examples include:

Social Media Engagement

The Wire has minimal social‑media presence, and its articles are rarely retweeted. Johnston discovered that roughly half of the site’s engagement on X (formerly Twitter) originates from Patrick Hynes, president of the PR firm Novus Public Affairs. Novus’s client list includes Targeted Victory, a consulting firm central to OpenAI’s lobbying efforts in Washington.

Broader Context of Generative AI

Generative AI is already blurring the line between reality and fabrication. With sufficient computing power, it can:

  • Create fake movie trailers that never existed – see an example from The Verge here.
  • Produce deep‑fake audio of politicians – see NPR’s coverage here.
  • Invent implausible scenarios, such as a shark attacking a plane – see the viral video here.

Implications

If Johnston’s reporting is accurate, The Wire may represent an instance where an AI‑driven firm mischaracterizes its output as “independent journalism” to further lobbying objectives. This would contravene OpenAI’s own usage policies.

Disclosure

Ziff Davis, Mashable’s parent company, filed a lawsuit against OpenAI in April, alleging that OpenAI infringed Ziff Davis copyrights in training and operating its AI systems.

0 views
Back to Blog

Related posts

Read more »

OpenAI available at FedRAMP Moderate

FedRAMP Moderate Authorization Overview OpenAI has achieved FedRAMP 20x Moderate authorizationhttps://www.fedramp.gov/marketplace/products/FR2533155773/ for Ch...