Quartz is testing a new kind of newsroom output: short AI-generated articles built from reporting published elsewhere. The international business news outlet has used the byline “Quartz Intelligence Newsroom” for these pieces, while its corporate parent, G/O Media, says the effort is still experimental.
The experiment matters because it shows how quickly automated publishing can move from narrow formats, such as earnings reports, into general news summaries. It also highlights a core tension for publishers: AI can produce fast, inexpensive copy, but readers still depend on clear sourcing, strong editing and useful detail.
What Quartz has been publishing
According to TechCrunch, Quartz began by publishing simple AI-generated earnings reports months ago. Beginning last week, the outlet expanded into short news articles.
As of Monday afternoon, TechCrunch reported that 18 AI-generated articles had been published. One example, titled “South Korea shares preliminary findings on Jeju Air crash investigation,” pulled together reporting from CNN, MSN and The Associated Press on MSN.com.
The pieces are not long. TechCrunch described each AI-generated article as roughly 400 words in length. They also do not include full quotes from sources, which changes how the articles read compared with conventional reporting.
The sourcing method is especially important. Instead of attributing details throughout the story, Quartz’s AI writer places source citations at the top of the article. That may tell readers where information came from, but it gives them less context about which outlet reported which fact inside the body of the piece.
G/O Media says the project is experimental
A spokesperson for G/O Media, Quartz’s corporate parent, confirmed to TechCrunch that the company has a “purely experimental” AI newsroom. The spokesperson did not say which AI models or tools are being used to produce the articles.
The company also did not make clear how Quartz’s AI newsroom decides which stories to cover. That leaves an important operational question unanswered: whether the system is choosing topics based on audience interest, available source material, editorial direction or some other process.
G/O Media’s stated goal is to free Quartz’s editorial staff to “work on longer and more deeply reported articles.” The spokesperson also said the editorial staff reviews each AI-generated story before publication.
That review process is central to the project’s credibility. If an AI newsroom is meant to handle shorter summaries, editors still need to make sure the output is accurate, specific and useful. The TechCrunch article argues that, at least in one case, the quality control did not appear strong enough.
Where the quality concerns show up
One example involved a TechCrunch how-to article about deleting Facebook, Instagram and Threads accounts. The original article included step-by-step instructions for downloading and saving data before deleting the accounts.
Quartz turned that source material into a much shorter AI-generated summary of about 300 words. Its headline, “How to delete your Facebook, Instagram, and Threads right now,” suggested a practical guide. But the instructions described by TechCrunch were broad rather than detailed.
The Quartz summary said Facebook users should go to “Settings & Privacy” and select “Account Ownership and Control.” It also said Instagram users could use the Account Center or settings to download data before deleting profiles, and that deleting Threads profiles requires removing the linked Instagram account because the two are interconnected.
For a reader trying to complete a task, that level of detail may not be enough. A how-to article has a different job from a news brief. If the headline promises immediate instructions, the body needs to help the reader act with confidence.
TechCrunch also pointed to another Quartz AI headline: “Jobless claims rise slightly as continuing claims set a record.” The criticism was that the wording created an awkward internal tension, with one measure described as only rising slightly while another was said to be setting a record.
Why publishers are interested in AI copy
The business appeal is straightforward. AI-generated content can give publishers a way to produce more articles with lower labor costs because AI does not require salary or benefits. For a media company, that can look like a path to more output and potentially more profit.
G/O Media told TechCrunch that reader response to and engagement with the AI stories have “far exceeded our expectations to this point.” The company also denied rumors of cash woes, saying it is “very well funded” and has a “good amount of working capital to draw on if needed.”
The spokesperson said previous staff reductions were connected to the sale of some sites in 2024. They also said Quartz is in the process of hiring more editorial staff.
Still, the experiment sits in a broader media context. G/O Media came under fire in July 2023 for publishing AI-generated content that contained errors and did not involve input from G/O’s editors or writers. Merrill Brown, the company’s editorial director at the time, defended the practice, while journalists at G/O-owned outlets such as Gizmodo objected.
The bigger issue for automated journalism
Quartz is not the only media organization to try AI-generated content. TechCrunch cited CNET and Gannett as publishers that have released factually inaccurate AI-generated stories and art. It also mentioned Sports Illustrated in connection with fabricated bylines.
The lesson is not only that AI can make mistakes. It is that automated publishing changes the burden on editors. If an article is assembled from other outlets’ reporting, readers need to know what came from where. If a piece claims to be useful, it needs enough detail to be useful. If a newsroom uses an AI byline, it still needs human accountability.
Quartz’s experiment may be intended to give journalists more time for deeper work. But the examples described by TechCrunch show why that promise depends on execution. Short AI articles can still affect reader trust, especially when they summarize reporting produced by other journalists.
For publishers, the question is no longer whether AI can generate a readable article. The harder question is whether the article gives readers enough sourcing, accuracy and practical value to deserve publication.