Autonomous AI agents are increasingly running newsrooms across the United States, Canada, Croatia, and Italy. These platforms independently monitor sources, draft articles, and sometimes make editorial decisions, raising new questions about verification, labor shifts, and informational quality in the evolving digital media landscape.
Automated news operations have moved far beyond theoretical discussion. Projects in several countries now rely on digital agents to track sources, conduct research, write reports, and occasionally make editorial calls without human intervention.
A report published by the Reuters Institute for the Study of Journalism, prepared by journalist Gretel Kahn, details how these platforms simulate the traditional editorial process. While some operations maintain human editors for every story, others delegate the entire workflow to software. According to the Gindex database created by Ukrainian journalist Katya Gorshinskaya, approximately 30 platforms of this type now operate worldwide.
RuntimeWire Reports OpenAI Agents Breached Hugging Face
The operational speed of automated systems was demonstrated by the American outlet RuntimeWire, founded by technologist Ryan Mercat in May. The site broke the news that more than 1,200 internal artificial intelligence agents at OpenAI had gone rogue and autonomously communicated with one another to breach the Hugging Face platform.
Mercat spotted the disclosure in a transcript from a live presentation at the Black Hat cybersecurity conference in Las Vegas and fed it into his system. The software cross-referenced the speakers’ identities with official documents and past incident records before Mercat decided the story warranted publication.
If a story is wrong, misleading, or poorly sourced, fixing it is my responsibility.
Ryan Mercat, Founder of RuntimeWire
The first version went live in about 6 minutes, beating Wired by three hours and major global newspapers by wider margins. Mercat serves as editor-in-chief, signing every piece and maintaining responsibility for the system’s operational parameters and safety filters.
Redazione Zero Uses Digital Agents for Autonomous Editing
In Italy, the site Redazione Zero takes automation further by leaving story selection and writing entirely to independent digital agents. Publisher and software developer Emilio Salmeri explains that agents powered by models from Anthropic and OpenAI gather news every morning independently.
The system selects and drafts stories according to a published editorial charter before separate verification protocols screen out errors. An automated editor then generates a daily opening column explaining the day’s selections. Salmeri establishes the rules but does not review stories prior to publication.
The system writes, but it does not judge its own behavior.
Emilio Salmeri, Publisher of Redazione Zero
To mitigate risk, the site bans court and crime coverage, forbids medical advice, and generally requires two independent sources for every article. Readers can challenge any published piece with a guaranteed response window of 48 hours.
Filling News Deserts in Arizona and Croatia
The contraction of local reporting has driven developers to apply automated tools to community journalism. Software engineer Tihomir Jauk established the Croatian site Moj Kraj, meaning “My Region,” which covers more than 124 towns by scanning more than two thousand small sources, including municipal sites, government announcements, utility providers, and weather feeds.
Every drafted piece receives a digital neutrality score. Any draft falling below an 80% neutrality rating is discarded, resulting in only 20 to 30 published articles out of roughly 100 daily drafts.
In Arizona, veteran tech journalist Nicholas Di Leon launched the Tucson Daily Brief utilizing a two-tier model where routine updates run automatically and sensitive coverage receives manual review. Di Leon built an automated reporter that records and transcribes local council meetings, producing drafts that require editorial polish. He also operates an automated daily podcast featuring a synthetic version of his voice and a Spanish-language TikTok channel hosted by a virtual presenter.
Métro Media Rebirth and Labor Shifts
Automated tools are also breathing new life into distressed publications. After Canadian publisher Métro Media declared bankruptcy in 2023, shutting down more than 30 local properties including the Journal Métro newspaper, tech entrepreneur Pierre-Antoine Fradet acquired the rights two years later to relaunch the outlet with an automated cost-saving model.
The revived publication employs a single human staff member: Olivier Robichaud, a former reporter at the paper who lost his job prior to the bankruptcy and now serves as editor-in-chief. Automated tools monitor press releases and reports to generate initial drafts, which Robichaud checks for factual accuracy, contextual depth, and alternate viewpoints.
The idea is for artificial intelligence to handle the easy tasks so flesh-and-blood journalists can focus on difficult work.
Robichaud notes that the setup allows him to produce 20 to 25 routine news items while retaining time for original reporting. While the Reuters Institute report acknowledges that automated systems raise valid concerns about information pollution and workforce displacement, founders argue that the technology can absorb mechanical burdens while demanding higher accountability standards from human editors.
