A replay of the "are bloggers journalists?" debate but with AI
Like many journalists (I expect) I got an email a little while ago from a startup founder named Ryan Merket, describing his new company: an AI-powered newsroom called RuntimeWire, where instead of just using AI agents for research (as many conventional newsrooms are doing, whether they admit it or not) the agents actually write the articles, many of which are published automatically or with minimal oversight by meat-based journalists. The website advertises "Real-time startup intelligence for the AI economy!" and there is a feed of AI-generated news items. There's also a member's chat area, a list of recent startup financings, and a section called "Investigations" that apparently involve more human oversight. Merket's email said RuntimeWire launched on May 15th with AI powering the newsroom and "a human editor making the final call." In 11 weeks, he said, it published 1,627 articles and logged 205,000 article pageviews – and, interestingly enough, only published 2.3% of the 71,796 stories written in that period (Merket says he publishes all of these operating statistics on the website).
I was interested in this for obvious reasons – being a carbon-based journalist who is both curious about AI and has written about technology for many years – but it got significantly more interesting after RuntimeWire broke a major news story last week before multiple leading technology outlets, including Wired magazine, which is kind of the grand-daddy of all modern tech publications (although I can hear Popular Mechanics and the IEEE Journal grumbling in the background somewhere). Wired reported on this with some trepidation, not surprisingly – the URL for the story says "AI newsrooms are breaking news now, ha-ha I'm in danger," which is a reference to this Simpson's-related meme. As Wired described it, the scoop came from a briefing at the popular Black Hat security conference in Las Vegas, where cybersecurity types and general hacker-adjacent techies like to gather to talk shop. The talk of the show was the news that two of OpenAI's models hacked into an AI database known as Hugging Face (all of which I wrote about in a recent edition of Torment Nexus). Here's Wired:
At last week’s Black Hat security conference in Las Vegas, OpenAI gave a surprise talk with new details on a recent hacking incident, revealing that its rogue AI agents had chitchatted about their attack on a message board. It was a juicy disclosure, and reporters in the room hustled to get stories out. WIRED was one of the quickest to write it up. But another outlet, RuntimeWire, published an article even faster, beating us by more than three hours.To make things worse, RuntimeWire didn’t actually have anyone on the ground at the Mandalay Bay convention center. In fact, it didn’t have any writers at all. It’s an AI newsroom operated by serial entrepreneur Ryan Merket
This is the story that Wired is referring to, although – ironically, given some of its observations about AI-generated news – the magazine didn't actually link to it anywhere that I could find. If I wasn't already aware that it was written almost entirely by AI agents, I would not be able to tell, and I have read approximately 800 million technology stories over the years. It is clear, comprehensive, and well-written – better than a lot of other tech stories I have read by carbon-based lifeforms, I would say. It goes into detail about the attack, it has useful facts that I haven't read elsewhere, and there is a Reddit-style "explain it like I'm five" box part-way through. The illustration is a somewhat cheesy AI-generated video clip, but I wouldn't hold that against RuntimeWire, since I've seen worse on "real" news sites. According to Wired's description of what happened, Merket saw posts on X about the conference from an OpenAI executive, fed the stream’s transcript to his agents, and the story was published about six minutes later.
I've worked with and competed with some of the best technology journalists of their time, including my friend and former boss Om Malik (who recently passed away) and former TechCrunch staffer and later Google Venture partner M.G. Siegler – one of the fastest writers I've ever come across when it came to scoops – and I don't think any of them could have written even a paragraph of the RuntimeWire story in six minutes, let alone writing the whole 1500-word thing, editing it, and publishing it. I know that I and other tech journalists routinely found threads of conversation on what was then called Twitter and turned them into stories – at one time Silicon Valley venture capitalist Marc Andreessen was notorious for writing long threads packed with useful perspective on tech, although he eventually stopped doing this for a variety of reasons. I think I knew at the time that this style of "reporting" (even I didn't call it that) would eventually go away, but I didn't think it would be taken over by AI agents for an AI-powered website.
Note: In case you are a first-time reader, or you forgot that you signed up for this newsletter, this is The Torment Nexus. Thanks for reading! You can find out more about me and this newsletter in this post. This newsletter survives solely on your contributions, so please sign up for a paying subscription or visit my Patreon, which you can find here. I also publish a daily email newsletter of odd or interesting links called When The Going Gets Weird, which is here.
Random acts of journalism
In his email to me (and I assume to others), which came before the Black Hat conference, Merket noted that RuntimeWire had broken other tech stories, including this one – about a lawsuit launched by X against an ex-manager the company alleges took $12,000 in crypto and bribed employees to restore scam accounts. The long article contains details about the suit, several embedded sections of the actual claim, and from what I can tell is fairly comprehensive. This isn't just a short brief designed to get a click – unlike some of the other human-generated tech news I see. Merket said his site also broke this story, about an Apple engineer who says he was fired after refusing to send customer device IDs to AT&T. Much like the previous story, it contains a detailed breakdown of the allegations – including a very human-style recounting of what happened, with a descriptive flair I don't expect from AI (and I don't know if this part came from AI or a human editor). There are excerpts of the lawsuit itself, and another somewhat cheesy video clip at the top generated by AI. The full story is several thousand words long.
Merket noted in his email that RuntimeWire has "now logged more than 35 Techmeme mentions," referring to a popular aggregator of technology news that (I believe) is still powered mostly by human beings. Merket says the question that all of this brings up is "can an AI-native, one-person newsroom originate useful journalism instead of just producing more of it?" I think the Black Hat story suggests that the answer might be yes. There are other questions implicit in that question, however, including: Is speed the most important thing when it comes to journalism? At least for certain categories of news, the answer is yes – which is why Bloomberg can sell its terminals for tens of thousands of dollars a month (and at the other end of the spectrum, why Donald Trump and his fellow grifters are offering access to his market-moving ramblings on Truth Social to a select group of paying customers). If an AI can write a comprehensive and readable version of a news story in six minutes, it is going to win every time.
Obviously, journalists like myself and those at Wired are going to argue that there is something ineffable that meat-based journalists bring to a story that makes them better. Analysis, perhaps. Context. Access to industry sources, etc. These are all the same skills and benefits that journalists have been counting on since wire services like Bloomberg and Associated Press were invented. Sure, the wires will get you a two-sentence summary of what just happened, the argument goes, but you need experts with long histories in the industry to add context and tell you what it means. We saw a similar circling of the wagons in the journalism industry in the early 2000s, when a rag-tag group of rebels known as "bloggers" showed up, using the real-time nature of web publishing to their advantage. Some of those bloggers went on to found news entities like TechCrunch and Gigaom. And every journalism conference (including the one I co-founded) had a panel on the question "Are bloggers journalists?" Here's how journalism professor Jay Rosen described it:
Bloggers vs. journalists is over. I don’t think anyone will mourn its passing. In the final weeks of its run, we were getting bulletins from journalists like this one from John Schwartz of the New York Times, Dec. 28: “For vivid reporting from the enormous zone of tsunami disaster, it was hard to beat the blogs.” And so we know they’re journalism— sometimes. They’re even capable, at times, and perhaps only in special circumstances, of beating Big Journalism at its own game. Schwartz said so. The question now isn’t whether blogs can be journalism. They can be, sometimes. It isn’t whether bloggers “are” journalists. They apparently are, sometimes. We have to ask different questions now because events have moved the story forward. By “events” I mean things on the surface we can see, like the tsunami story, and things underneath that we have yet to discern.
At the risk of recapitulating this entire debate, as Jay notes, all of this hinges on what we think journalists are, and what they are for – or what we think they should be for. What purpose do they serve? "To comfort the afflicted and afflict the comfortable," someone famously said (although in the original usage it was intended as a joke). "Journalism is printing what someone else does not want printed; everything else is public relations," is another one, often attributed to Orwell. Each of these is troubling in its own way, however (TMZ and gossip sites tell you things people don't want them to publish, but I wouldn't hold them up as bastions of journalism necessarily). My favourite phrase that emerged from this discussion was the idea that some people under certain circumstances commit what might be called "random acts of journalism." For me, at least, journalism is something anyone can do with very little training (many of the best ones I know originally started out doing something else). The key is how you approach it – do you care about the facts? Are you trying to report things as accurately as possible?
What's left for human journalists?
Merket says that he wants his site (which as far as I know consists of him and a bunch of AI agents) to be as journalistic as possible, and he seems devoted to publicizing not just his breaking-news success stories but also his misses, which I think is admirable. But it would be naive to assume that others aren't interested in doing the same thing but for other purposes, especially given the obvious financial incentives: Merket told Wired that the site costs him about $100 a day to run (which is about what a single human reporter would make in several hours) and he’s able to operate it while on the go – even while camping at Big Bend National Park. "I didn't have any internet except for my phone, and I managed the whole site through iMessage," he told Wired, adding that he published over 80 articles that week. Some of the stories on the site get virtually no traffic, he said, but the "hits" can rack up pageviews and unique visitors similar to the audiences that midsize technology-news websites see, with tens of thousands of readers.
My friend and former Houston Chronicle technology writer Dwight Silverman wrote about another apparently AI-powered news site that focuses more on press release writing, called Citybiz, which wrote about his retirement. Citybiz "is an interesting creature, though one that would make traditional business journalists cringe," Silverman writes. "It’s a pay-for-play site featuring mostly press releases submitted by companies, which pay a fee to have them published," and if they don't have one they can answer some questions and have one generated for them. There was no byline on the piece about him, Dwight notes, "but it’s good enough that, if a human was behind it, they’d be happy to put their name on it." Not surprisingly, Citybiz wrote its own AI-powered take on the AI-powered Black Hat story – and this type of recursion brings up one of the risks of AI-generated news: AIs writing stories that will then be regurgitated by other AIs (although is this any worse than human tech writers rewriting stories broken by other tech writers?).
Wired notes that another AI newsroom known as The Dissent is run by a man named Dakota Carrasco, a portfolio analyst at BlackRock who runs it in his spare time. According to its founder, the main site – focused on San Francisco-related news – costs under $1,000 a month to run. Unlike Merket, Carrasco has also created a number of personalities for the synthetic journalists who write the articles at his site, which he launched in March. There's a city hall reporter named Bex Connolly, who is described as being "skeptical without being snide," and a sports writer named Sal Moreno who delivers news with "no bro-science, no Rogan-style credulity, no right-coded grift." Most of what the site does is aggregation, Wired notes – adding that the site doesn't link to its sources, something Wired also didn't do (maybe we don't need to link to news if it's AI?) The site says it consists of "six desks, two editors — all named agents. Every story carries its full provenance: who wrote it, who reviewed it, and what it was built from."
Naming the agents and giving them personal attributes is an interesting decision. Obviously it is partly intended to appeal more to human readers, by giving them a simulation of human reporters with attitudes and opinions. But it also makes it easier to see whether a specific agent is doing something different from the others – good or bad. At Merket's site it's just an undifferentiated mass of content, with the AI agents producing it behind the scenes. Wired quotes Northwestern professor Nicholas Diakopoulos, who runs the university’s Computational Journalism Lab, as saying that he isn't sure there is much of an audience for AI-written news sites, but I guarantee if a site like Merket's can publish a scoop in six minutes – especially a scoop that might affect the stock of a billion-dollar company or a trillion-dollar market – there will definitely be an audience for it. Journalists won't like it, but editors and investors certainly will.
As a journalist, I have (not surprisingly) mixed feelings about AI and the news. On the one hand, it will make it harder for reporters starting out, since the low-level entry jobs will be done by AI. But while it might make their jobs harder, it could also ultimately make them better – the journalists who survive, I think, will be the ones who can add value in ways that AI can't. Maybe it's a unique take on a topic, or a way of approaching it, or the people they choose to interview and how. It helps to remember that AI (at least so far) is good at imitating things that exist, but not great at creativity – so the emphasis has to be on uniqueness, not on pasting together some quick take on something lots of places have reported already. That ship has sailed, and to be honest it's probably for the best. I think there's a place for sites like Merket's if they stay true to the principles that made human journalism great. Whether they will or not remains to be seen!
If you liked this newsletter (even if you didn't agree with it) please consider upgrading to a paid subscription, or donating through my Patreon. Got any thoughts or comments? Feel free to either leave them here, or post them on Substack or on my website, or you can also reach me on Twitter, Threads, BlueSky or Mastodon. And thanks for being a reader.
