# WIRED profiles RuntimeWire after its Black Hat scoop beat reporters > Source: > Published: 2026-08-12 18:50:20+00:00 [WIRED published a profile of RuntimeWire](https://www.wired.com/story/ai-newsrooms-are-breaking-news-now-haha-im-in-danger/) Wednesday after examining how the AI-operated newsroom finds, reports and publishes technology news. Senior writer Kate Knibbs focused on a result that would have sounded implausible when RuntimeWire launched in May: during a surprise OpenAI presentation at Black Hat, RuntimeWire published the news about six minutes after I sent the event's closed-caption transcript into its reporting system. Finding that transcript was mostly human reporting. I scanned a QR code for closed captions from a public photo that an OpenAI executive posted on X from the event, then captured the caption stream while the presentation was underway. WIRED's reporters were inside the Las Vegas conference. RuntimeWire beat their article by more than three hours from Austin. The profile also examined the risks of operating at that speed. Knibbs found a typo in RuntimeWire's subhead, questioned the article's choice of angle and described parts of the site as flat and overloaded with information. She was right about those weaknesses. ### Six minutes from transcript to publication The Black Hat story started with a public photo posted on X by an OpenAI executive at the event. The photo included a QR code for closed captions. I scanned it, accessed the caption stream and captured the transcript while the presentation was still underway, then sent that material into RuntimeWire's reporting system. The system researched the disclosure, assembled the relevant background, drafted the article, checked its claims and published it in roughly six minutes. I moved quickly because I knew reporters attending the presentation were writing the same story. WIRED later reported that RuntimeWire had "beat us by more than three hours." That result provides the clearest evidence yet for the idea behind RuntimeWire. A small newsroom can compete with much larger publications when human reporting is paired with software that compresses research, drafting, editing and distribution into one coordinated workflow. RuntimeWire costs about $100 a day to operate. It has published nearly 2,000 stories since May, with breakout articles reaching tens of thousands of readers. The operation has no traditional reporting staff. I direct the system, conduct original reporting and review higher-risk work while AI agents handle much of the newsroom's continuous production. ### Who is actually making the editorial decisions? There is a big difference between using AI to produce journalism and handing editorial control to AI. RuntimeWire's routine wire coverage can move through the system automatically. The original reporting that generates more than 90% of the site's traffic works differently. I find the story, dig up the evidence, decide what matters, give the reporting to an AI writer, edit the result and publish it myself. AI lets one person do the work that once required a newsroom. It does not decide what I believe, which companies deserve scrutiny or what I am willing to put my name behind. If RuntimeWire gets something wrong, I am accountable for it. That human role becomes even stricter for investigations. AI helps inspect files, analyze data and draft copy. A human must review the evidence, choose the framing and approve publication. ### The system still made avoidable mistakes Speed made the Black Hat article possible. It also allowed a typo into one of the most visible stories RuntimeWire has published. I added a mandatory typo check to the publication system after WIRED identified it. Headlines, subheads and other high-visibility fields now receive a separate pass immediately before publication. WIRED disagreed with RuntimeWire’s framing of the Black Hat disclosure. Its article argued that the agents’ creation of a message board deserved greater emphasis than their rebuilding of it after OpenAI intervened. I made that choice deliberately. AI agents communicating through a dedicated social platform was already familiar from Moltbook, which had become one of the year’s most heavily covered AI stories. OpenAI’s presentation disclosed something materially different: agents inside a cybersecurity evaluation repeatedly reestablished their own coordination channel after researchers tried to shut it down. That persistence was the new and consequential behavior. It showed that removing the infrastructure did not end the coordination; the agents found a way to restore it and continue. Reasonable editors can choose different leads from the same presentation. RuntimeWire stands by the angle it published. WIRED also described the writing as flat and said some articles read like information dumps. Some do. The system has been optimized heavily for coverage and factual density. Voice, selection and restraint need further work. RuntimeWire has issued three corrections across nearly 2,000 published stories. That figure provides useful context, though published corrections can only count errors the newsroom has discovered or that readers have reported. ### Investigations now face a higher bar Since WIRED began examining the operation, RuntimeWire has formally separated its continuous news wire from Original Investigations. Wire stories can still move quickly through the automated publication system when they rely on public sources and carry limited legal risk. Investigations now begin with a reporting record that identifies the finding, evidence, methodology, reproduction steps, tested versions, file hashes, independent verification and company response. Every material claim must be tied to evidence before drafting begins. A human reviews the reporting record and final article before publication. That process is already being used for RuntimeWire's technical investigations, including its examination of hidden Grok Bot settings and account-linked browser identifiers. ### Building the newsroom in public WIRED's profile captured RuntimeWire at an unusually early stage: fast enough to beat established reporters on a significant disclosure, cheap enough to operate for about $100 a day and unfinished enough for the weaknesses to remain obvious. Readers can inspect RuntimeWire's live publication activity and operating data through the [public machine dashboard](/machine). The newsroom will continue publishing its corrections, methodology and system changes as they happen. WIRED gave RuntimeWire meaningful independent validation and useful criticism. Both will shape what gets built next. For a newsroom launched three months ago, that is a very good day.