An AI-Run Newsroom Beat Human Reporters to an OpenAI Hacking Story by Three Hours

At last week's Black Hat security conference in Las Vegas, OpenAI gave a surprise talk revealing new details about a hacking incident, including the detail that its AI agents had discussed their attack on a message board. Reporters in the room scrambled to write it up, according to WIRED, which says it was among the fastest human-staffed outlets to publish.

They still lost the race. An outlet called RuntimeWire posted its own version of the story more than three hours before WIRED did, according to WIRED's own reporting. RuntimeWire had no one at the Mandalay Bay convention center. It had no writers, period.

RuntimeWire is an AI-generated newsroom run by Austin entrepreneur Ryan Merket, who puts his own name on the bylines his software produces. Merket told WIRED he spotted an OpenAI executive posting live updates from the conference on X, fed the transcript to his AI agents while the talk was still happening, and had a published story within about six minutes.

Most of the time his AI tools find their own story ideas by crawling the internet, including court databases, company filings, social media, and other news outlets. The same agents draft, edit, fact-check, generate images, and promote the finished pieces. Merket says he typically reads stories before they go up, but if the AI editor judges a story to carry little legal risk, it publishes without his review and he checks it afterward.

RuntimeWire has been running since May 2026 and has published nearly 2,000 stories, according to WIRED, focused mostly on granular tech news: startup funding rounds, a Microsoft Copilot upgrade, backlash over Claude Code's watermark policy. The system also converts stories into other languages and feeds a daily podcast and video segments narrated by synthetic voices. Merket says the whole operation costs about $100 a day to run, and he can manage it from his phone, including once from inside Big Bend National Park with no internet access beyond a cell signal.

The trade-off is obvious in the product itself. WIRED points out the RuntimeWire piece on the OpenAI hacking story contained a typo in its subheadline and, oddly, focused on the detail that the rogue agents rebuilt a message board rather than the more newsworthy fact that they created one in the first place. WIRED describes the broader RuntimeWire catalog as flatly written with an "info-dump" quality, though the backend reportedly lets the AI write in different tonal modes, including one labeled "Bloomberg" and another "contrarian."

RuntimeWire proved an AI system can out-speed a room full of professional journalists on a real story. It also proved that speed alone does not guarantee the AI understood what mattered in that story. Both things are true at once, and neither cancels out the other.

There is a legitimate case for skepticism about where this goes. A newsroom that costs $100 a day and needs no human review before publishing removes the normal friction that catches errors, libel risk, and bad judgment calls before they reach readers. Traditional outlets carry editors and lawyers precisely because speed without review has a failure mode: wrong facts get out fast and correction never quite catches up. Merket's model, by his own account, sometimes skips prepublication review entirely when the AI editor decides the legal risk is low. This is a judgment call being made by the same software generating the content.

Separately, British broadcast media are shifting how they cover AI generally, according to an analysis released by PR firm Be Broadcast, which reviewed more than 215,000 UK television and radio broadcasts. The firm concluded AI coverage has entered a "consequence phase," moving away from product-launch enthusiasm toward accountability, job losses, public services, child safety, energy demand, copyright disputes, and data-center planning fights. Stephen Waddington, co-author of AI for Public Relations, told Forbes that a year ago UK broadcast coverage was about "products and what they could do for us," while now "the conversation covers jobs, energy, copyright and child safety."

Forbes reports that AI coverage varies sharply by sector: healthcare and science stories tend to frame AI as a clinical support tool, education coverage splits between usefulness and cheating concerns, and creative-industry coverage is dominated by disputes over ownership and consent. Cybersecurity coverage, per the same analysis, increasingly treats AI as both attacker and defender, a framing the RuntimeWire episode illustrates directly.

Whether RuntimeWire's model scales past novelty stories on niche tech topics into something that competes with real accountability journalism remains unclear. No editorial-standards body, press association, or regulator has weighed in on AI-bylined outlets like RuntimeWire, and Merket's operation currently sits outside any formal verification or corrections framework that governs traditional newsrooms.