In April last year, I wrote what amounted to an obituary for Quartz, the innovative digital business publication that had once promised to reinvent journalism. I'd followed its development since before its launch in 2012, interviewing its founding editor and later its publisher for the For Immediate Release podcast.
I admired what Quartz was trying to achieve. Its eventual decline was a sad reminder that even brilliant editorial ideas need a viable business model. As co-founder Zach Seward acknowledged in his own farewell to the publication, Quartz had never made money.
Except Quartz didn't quite disappear. The name survived, as did its website. And according to a new investigation by Press Gazette, what appears on that website today bears little resemblance to the publication I once admired.
One reporter, apparently using AI extensively, produced more than 50 stories in a single day. That's an extraordinary volume of content for any journalist. It's also a striking development in the story of a publication that once wanted to escape the pageview-driven economics of digital media.
So what happened? And what might Quartz's strange afterlife tell us about the future of journalism?
From innovation to automation
When Quartz launched in 2012, it was genuinely different. Designed for mobile devices before mobile-first publishing became commonplace, it combined elegant design, an uncluttered reading experience and an editorial approach that treated readers as intelligent participants rather than targets for advertising.
I remember interviewing founding editor Kevin Delaney a few months before launch. He described a newsroom designed from scratch for a digital-first world, free of the legacy thinking that constrained traditional publishers. When I spoke with publisher Jay Lauf two years later, he explained how Quartz approached publishing as a product, not simply a collection of articles.
Its distinctive editorial approach, including its famous obsessions instead of conventional news beats, was part of a broader attempt to rethink how people discovered and understood business news.
The ambition was to create something valuable enough to earn readers' attention without chasing clicks for their own sake.
Fourteen years later, Quartz appears to be pursuing a very different kind of innovation.
In his 9 October report for Press Gazette, Rob Waugh describes how a reporter named Cris Tolomia produced at least 52 articles for Quartz on 8 October alone. The output, estimated at around 20,000 words a day, ranges across subjects as diverse as stock markets, energy, cybersecurity, AI drug research and the best dog breeds for children.
According to the investigation, much of this material consists of rewrites of reporting from other sources, apparently produced with extensive AI assistance. Tests using an AI-detection tool classified two examined articles as entirely AI-generated and another as predominantly AI-generated.
Such detection tools cannot conclusively establish how an article was written, and publishing volume alone tells us little about accuracy. But the sheer scale of production raises obvious questions about what journalism involves when one person's byline appears on dozens of stories every day.
It's difficult to imagine a greater contrast with the editorial philosophy that originally distinguished Quartz.
Two publishing models under one roof
However, there's an interesting complication. The current Quartz isn't necessarily abandoning original journalism altogether.
According to Press Gazette, Quartz's newsletters still appear to offer original journalism. The publication has also launched Power Play, a six-part investigative podcast exploring the growing demands AI is placing on America's electricity grid.
That suggests something potentially significant: two quite different approaches to publishing operating under the same familiar brand.
On one side are newsletters, specialist reporting and podcasts, where the value comes from originality, editorial judgement and a direct relationship with readers.
On the other is a website apparently producing large quantities of inexpensive, AI-assisted articles, many based on information already published elsewhere. Some might call this 'AI slop'. But is that a fair description, or does the real problem lie in what this approach says about the value of journalism?
Perhaps this is a deliberate strategy to make the economics work. If automation makes routine website publishing much cheaper, the resulting revenue might help support more ambitious original journalism.
That's not an unreasonable proposition. News organisations have struggled for years with declining advertising income, changing reader habits and the growing difficulty of attracting traffic from search engines and social platforms.
But there's another possibility. Filling a once-respected publication's website with derivative, largely automated content could undermine the very reputation that makes its original journalism valuable.
I don't yet know enough about Quartz's commercial performance or internal editorial processes to judge which outcome is more likely.
What I can say is that the contrast is remarkable. Quartz once tried to solve digital publishing's economic problems by making the journalism more distinctive. Its website now appears to be approaching the problem by making content production dramatically more efficient.
When more content doesn't mean more journalism
There's nothing inherently wrong with using AI to make journalism more efficient. I use AI in my own writing and research, and I've argued before that the presence of AI in the creative process matters less than the quality and integrity of the finished work.
AI can help journalists analyse documents, identify patterns in data, transcribe interviews, organise research and develop drafts. Used thoughtfully, these capabilities could give journalists more time to investigate stories, challenge assumptions and produce original reporting.
For smaller publishers with limited resources, that could be transformative.
But there's an important distinction between helping journalists do their work and using AI primarily to produce more content.
Rewriting someone else's reporting into a new article may be useful to readers, but it doesn't necessarily add anything to our collective understanding of the subject.
And if producing content becomes almost infinitely scalable, what happens to the economic value of the journalism that makes it possible?
Consider the potential consequences. A publisher invests time and money in original reporting. AI then summarises, rewrites, and republishes the story across numerous other websites. Those derivative articles may compete for the same readers and advertising revenue, without bearing the original reporting costs.
If that becomes the dominant model, who will pay for the reporting everyone else depends on?
The danger isn't that AI will make writing easier. It's that the economics of automated publishing could reward reproduction more generously than discovery.
The real question isn't whether AI wrote it
Quartz is unusually open about AI's role in its operations.
Its editorial policy states that AI is central to how it reports, produces and publishes news. It distinguishes original journalism, which it says is entirely reported, written and edited by journalists, from routine coverage and other material that may be AI-assisted or AI-driven.
It also promises human editorial oversight throughout.
What particularly caught my attention was Quartz's explanation for why it doesn't label individual stories by their use of AI. It compares the choice to the way publishers don't identify which word processor a journalist used to write an article.
I understand the argument, and to a point I agree with it.
I've never believed that readers need a detailed inventory of every tool used to create a piece of writing. If AI helped a journalist structure an article, improve the prose or make sense of a complicated document, that doesn't automatically diminish the work.
But the word processor comparison only goes so far.
A word processor doesn't independently generate reporting, rewrite other people's articles, or potentially introduce factual claims that its user hasn't verified. Generative AI can do all those things.
That doesn't make its use unacceptable. It makes human involvement more important.
These are questions about the publishing process, not accusations against the person whose name appears on the articles. We don't know precisely how Quartz divides the work between AI systems, reporters and editors.
Press Gazette says it sought comment from Quartz and Redbrick, its current owner, but received no response before publication.
I'd be interested to hear the explanation. Not because using AI to produce news is inherently objectionable, but because the meaning of a journalistic byline is changing.
A byline has traditionally signalled authorship and responsibility. In an AI-assisted newsroom, it might increasingly identify the person supervising a production process rather than personally researching and writing the resulting article.
That's a significant change, and one that deserves more transparency than a general declaration that humans remain involved.
What kind of publishing future are we creating?
The Quartz story arrives at an awkward moment for journalism.
AI is making content production cheaper and faster just as AI-powered search and answer engines are changing how people discover and consume information. Increasingly, people can get answers without visiting the publishers whose work contributed to them.
Publishers therefore face pressure from both directions: producing original journalism remains expensive, while the traditional routes to earning revenue from it are becoming less dependable.
Automation offers an obvious response. Reduce production costs, increase output and find ways to monetise whatever audience remains.
I think there is a more promising possibility.
AI could take over some of the routine work that consumes journalists' time, while publishers invest more heavily in investigation, expertise, analysis and original perspectives. Newsletters and podcasts could strengthen relationships with readers who value those qualities enough to seek them out.
Perhaps that is partly what Quartz's current owners are trying to achieve.
But technology alone won't determine the result. Publishers will make commercial and editorial choices about where the savings go, what standards they maintain and what kind of relationship they want with their audiences.
If the overriding objective becomes producing the greatest volume of content at the lowest possible cost, we risk creating a publishing environment overflowing with information but increasingly short of original journalism.
What made Quartz worth reading
When I wrote my Quartz obituary last year, I ended by thanking the publication for showing us a glimpse of what digital journalism could be.
That original vision wasn't about producing more articles, faster or cheaper. It was about creating journalism that respected readers' intelligence, offered fresh perspectives and helped people make sense of a complicated world.
AI may yet help publishers achieve some of those ambitions in ways the founders of Quartz could scarcely have imagined.
Producing 50 articles a day shows extraordinary publishing capacity.
Whether that translates into journalism worth reading is an entirely different question.
Principal sources:
- Quartz website mainly produced by one AI-assisted reporter writing 50+ stories per day (Press Gazette, 9 October 2026)
- Quartz's current editorial policy, including its AI disclosure (About Quartz)
- Quartz: From Bold Vision to Obituary (Studio6, 10 April 2025)
(Screenshot of the Quartz homepage captured by the author.)
