Skip to content
Article Reuters Institute for the Study of Journalism Oct 2026

Reuters Institute: inside the rise of AI-native newsrooms run by autonomous agents

By October 2026, Jindex.ai — a database maintained by journalist and researcher Katya Gorchinskaya — had logged around 30 AI-native news outlets worldwide. Reuters Institute journalist Gretel Kahn spoke with five of them: RuntimeWire (US), Redazione Zero (Italy), Moj Kraj (Croatia), Tucson Daily Brief (US), and Journal Métro (Canada). The resulting piece, published on October 16, 2026, reads less as a verdict on the category and more as a close study of the choices each outlet made about where humans fit.

What “AI-native” covers

The outlets differ significantly. RuntimeWire’s editor-in-chief, Ryan Merket, approves every story before publication and signs each one. His AI pipeline — which ingested a conference transcript and produced a publishable article six minutes later — is fast infrastructure with a human at the end. Redazione Zero goes further in the other direction: separate AI agents gather, write, and fact-check stories, with a final AI editor producing the daily editorial and no human approving stories before they go live. Journal Métro sits between them: AI drafts stories from trusted sources, and editor-in-chief Olivier Robichaud verifies each one.

What the guardrails look like

Each outlet describes structured safeguards. Redazione Zero bans crime and court reporting from its automated pipeline and requires two confirmed sources per story; it has issued four corrections since launching, and each led to new constraints on the system. Tucson Daily Brief uses automatic publishing for routine items and manual review for higher-stakes stories. Moj Kraj scores every draft for neutrality and other criteria before it publishes, running roughly 100 drafts daily and publishing 20 to 30.

Tucson Daily Brief’s founder, Nicholas De Leon, notes that his AI “live reporter” drafts town hall coverage that is about 90% publication-ready. The outlet also runs an AI-voiced podcast, a Spanish-language TikTok channel, and an AI-based search tool.

The honest accounting

The outlets are small, most are not yet profitable, and several operate in coverage gaps — hyperlocal beats, regional markets — where they are not competing directly with established newsrooms. Their founders generally argue that AI should handle aggregation and routine monitoring so journalists can focus on reporting that requires sources and fieldwork.

Whether that framing holds as the category grows is an open question. Robichaud argues that job outcomes depend on what employers choose to do, not on what the technology makes possible.

Who it’s for

The article is useful for editors and journalists evaluating autonomous editorial workflows, for newsroom leaders setting AI policy, and for anyone following how different human-oversight models perform in practice.