Where we came from
AI Frontier Review was founded in late 2023, in the months after the field's transition from research curiosity to industrial build-out. The desk began as a small group of reporters and editors who had been covering frontier AI in adjacent beats — research, infrastructure, lab strategy — and had grown impatient with how the story was being told elsewhere.
The early coverage of frontier AI suffered, in our view, from three failure modes. The capability-demo writeup, in which a model's most impressive output got mistaken for its typical one. The doom-laden think piece, in which the absence of a specific source was compensated for with the height of the rhetoric. And the press release rewritten as news, in which a lab's framing of its own work became, with minimal interrogation, the frame readers received.
We started with three reporters and a working list of labs we wanted to cover. The list has grown. The newsroom has grown. Most of what we publish is news — short, dated, bylined, and sourced. We also file features and opinion, both clearly labeled.
Our standards
Every story on AI Frontier Review is reported by a named member of the desk. We do not run editorial-we copy. We do not pseudonymously aggregate. Every factual claim in a story we publish traces to a primary source — a paper, a release, an official filing, an interview that actually happened. When we quote a person, the quote is verbatim and the source is linked at the bottom of the story.
When the desk cannot verify a story to that standard, the story does not run. We have killed pieces we would have liked to publish, and we expect to kill more. The alternative — putting unverified claims under our masthead — costs us more than the speed it would save.
Corrections and tips
Corrections appear inline on the story they correct, dated and explained. Tip and document submissions go through the newsroom, where each reporter is listed alongside the beat they cover. The desk treats material shared in confidence as confidential, and verifies every claim against an independent source before publication.
Subscribing
A full RSS feed carries every story by date. A sitemap lists every page on the site for crawlers and readers alike.