AI Newsroom Breaks Stories in 6 Minutes With No Reporters

Ryan Merket beat some of the biggest tech reporters in the world to a major story last week, and he wasn't even at the conference where it broke.

The Austin entrepreneur runs RuntimeWire, a newsroom with no human writers. His AI agents find stories, write them, fact-check them, create images, and publish them, often without him reading them first. When OpenAI revealed details about a hacking incident at a Las Vegas security conference, Merket fed his AI the conference transcript from home and published a complete article in six minutes.

RuntimeWire has published nearly 2,000 stories since launching in May. The AI tools crawl court databases, web forums, company filings, and social media to find tech news worth covering. Recent stories include biotech startup funding rounds, software updates, and tech policy debates.

The operation costs about $100 per day to run. Merket manages the entire site from his phone, even publishing over 80 articles in one week while camping in Big Bend National Park with spotty internet. He uses iMessage to oversee his digital newsroom from remote locations.

The stories that hit can reach tens of thousands of readers, comparable to midsize tech websites. One of Merket's childhood friends already shared a RuntimeWire story without realizing it was AI-generated. The misses barely get any traffic, but the overhead is so low that it doesn't matter.

Merket isn't alone in this experiment. Dakota Carrasco, a BlackRock analyst, runs a similar operation called The Dissent focused on San Francisco news. His AI newsroom costs under $1,000 monthly and features synthetic journalists with distinct personalities, like skeptical City Hall reporter Bex Connolly and sports writer Sal Moreno.

The Ripple Effect

These automated newsrooms represent a shift in how breaking news might work. Traditional outlets need reporters on the ground, editors reviewing copy, and significant overhead. AI newsrooms can operate anywhere with an internet connection and respond to breaking news in minutes instead of hours.

The quality still trails human reporting. RuntimeWire's OpenAI story had typos and awkward phrasing. But the speed advantage is undeniable, and the technology improves constantly.

Merket recently split his newsroom into two sections to distinguish fully automated stories from those involving human oversight. Both still use AI for drafting. An AI agent assigns each story a legal risk score, and only low-risk pieces get auto-published.

The experiment proves that AI can break news, not just repackage it. Whether that's progress depends on whether speed matters more than the human judgment that catches what machines miss.