An AI Newsroom Beat Reporters Who Were in the Room
RuntimeWire published an OpenAI security story three hours ahead of journalists attending the conference in person. One human, a pipeline of agents, and about 1,600 articles since May.
RuntimeWire is a technology news site run by one person and a stack of AI agents. This month it published details from an OpenAI cybersecurity presentation at the Black Hat conference more than three hours before some of the journalists who were physically in the room filed their stories. Founder Ryan Merket spotted an OpenAI executive posting about the talk on X, fed the live stream transcript to his agents while the session was still running, and had a publishable article within minutes.
The operation is small by any traditional measure and large by output. Since launching in May, RuntimeWire says it has scanned 71,796 potential stories, published about 2.3 percent of them, and put out roughly 1,600 to 2,000 articles drawing around 205,000 pageviews. The pipeline identifies candidate stories, researches them, drafts copy, edits, fact checks, generates images, translates, and produces audio and video versions. Merket puts his name on the bylines and reviews higher risk pieces himself. Lower risk stories can go live before he has read them.
It is worth being precise about what happened and what did not. This was not investigative reporting. It was a public livestream, a transcript, and speed. The agents did not develop a source, verify a leaked document or push back on a company’s framing. What they did was collapse the distance between “information became public” and “an article exists”, and they did it while credentialed reporters were still walking out of a conference hall. That is a narrower achievement than the headline suggests and a more disruptive one than it sounds, because the economics of a lot of daily news coverage rest exactly on that gap. Monitoring court records, company filings, conference streams and technical forums is expensive labour that AI can do continuously, which means a two person outfit can now cover a beat that used to need a newsroom.
The obvious worries stand. Accuracy, sourcing, accountability and copyright do not solve themselves when the writer is software, and “publish before a human reads it” is a policy that will eventually produce a correction nobody catches. Speed on its own is also not obviously good for readers; the internet is not short of fast, and the things it is short of are verification and judgment. But the competitive pressure is real regardless of whether you approve.
What this means for you: be a bit more careful about where your AI and tech news comes from. Sites that are entirely or mostly machine generated are now good enough that you will not spot them by tone alone, so look for a masthead, named reporters, a corrections policy and original sourcing rather than rewrites. If you run a business that publishes anything, the useful takeaway is not “replace the writers” but “notice what part of the job was actually speed”. Monitoring, summarising and first drafts are now cheap. Judgment, relationships and being right are not, and those are the parts worth paying for.
Sources
Source: https://gizmodo.com/an-ai-powered-news-site-scooped-human-journalists-now-what-2000797885
One Researcher, Under 20 Prompts, One Day: How AI Built a Zoom Attack
A Security disclosed three critical Zoom flaws on August 11. The bug is bad enough, but the finding is the cost: work that used to take nation states months took a single researcher a day.
