OpenAI adds Folha de S.Paulo and UOL to ChatGPT's answer surface in Brazil

What actually changes on screen

The concrete deliverable in this announcement is a display change. OpenAI says ChatGPT users will be able to "access high-quality journalism and see summaries based on Folha de S.Paulo and UOL's reporting." That is a frontend claim, not a backend one — the value is in what renders in the answer, how it is labeled, and where the link points.

OpenAI frames the integration around three surface-level properties: "attribution, transparency, and links back to original sources." Each of those is a UI decision. Attribution means a visible publisher name attached to a summary. Transparency means the user can tell the answer draws on Folha or UOL rather than an unsourced blend. Links back mean the answer is a doorway, not a terminus — the reader can click through to the original article.

For anyone building AI answer experiences, this is the interesting part: the partnership is defined by how content is presented in the response, not merely whether the model can produce it.

The scale that makes the answer surface matter

The announcement puts specific numbers behind the reach. OpenAI cites "more than 900 million weekly active ChatGPT users around the world" as the audience for these summaries, and describes Brazil as "one of the largest markets for ChatGPT globally, with more than 50 million monthly active users and around 140 million messages exchanged per day."

Those figures explain why a display integration is worth a formal partnership. At 140 million messages a day in one country, the placement of a publisher's name and a clickable source link inside answers becomes a distribution channel in its own right. The frontend real estate — the few lines of a summary and the citation beneath it — is the product being negotiated.

By making their journalism accessible in ChatGPT, we want to bring more useful, timely, and locally relevant answers to ChatGPT, while supporting the broader news ecosystem.Montana Labs

Two audiences, two different frontends

The deal has a second, easily overlooked component. Beyond appearing in consumer ChatGPT answers, Grupo Folha and Grupo UOL "will also gain access to Codex, ChatGPT Enterprise, and API to explore new ways AI can support their journalism, develop innovative products and features for readers, and enhance internal workflows and business operations."

So there are two distinct surfaces in play. One is the reader-facing ChatGPT answer, controlled by OpenAI, where Folha and UOL content shows up as attributed summaries. The other is the publisher-facing tooling — Codex, Enterprise, the API — that the newsrooms themselves can build against to create their own reader features. The first is a placement the publishers receive; the second is a toolkit they wield.

Carlos Ponce de Leon, Folha's Co-CEO, frames the second surface as the strategic bet: "we are putting Folha at the forefront of this transformation and creating new ways to expand the reach, relevance, and impact of trusted journalism." That is the ambition to build, not just be cited.

The implication: attribution UI becomes the negotiated deliverable

The signal in this Brazil-first partnership — following deals OpenAI cites in the US, UK, France, and Germany — is that citation design is being treated as a first-class product feature, one that publishers now sign contracts over. The commitment to "attribution, transparency, and links back to original sources" is a promise about pixels: what the user sees crediting the source, and whether that credit is clickable.

For teams building AI-powered reading and answer experiences, the practical lesson is that the trust layer lives in the frontend. A summary without a visible source is a liability; the same summary with a labeled attribution and a working link back is a distribution asset for the publisher and a credibility signal for the reader. This announcement makes that surface the thing worth partnering over.