They are betting that collaboration is better than resistance
In May 2026, two of Brazil's most influential media institutions — Folha de S.Paulo and UOL — chose negotiation over resistance, becoming the first major Latin American publishers to formally license their journalism to OpenAI for use in training ChatGPT. The agreement arrives at a moment when the global press is slowly rewriting its relationship with artificial intelligence, moving from silent extraction toward something closer to mutual recognition. It is, at its core, a question as old as publishing itself: who owns the words, and what are they worth.
- For years, AI companies harvested journalistic work without consent or compensation, leaving newsrooms watching their reporting disappear into training datasets with no recourse.
- Folha de S.Paulo and UOL — institutions that together shape how millions of Brazilians understand their world — broke from regional inertia by formalizing a licensing deal that puts a price on their archives.
- The undisclosed financial terms mirror OpenAI's global playbook: legal access to content in exchange for compensation and some degree of attribution control, a structure that trades sovereignty for sustainability.
- Across Latin America, publishers in Mexico, Argentina, and Colombia now face a quiet ultimatum — negotiate your own terms or risk being bypassed entirely as AI systems grow more capable and less dependent on any single source.
- The deal lands not as a triumph but as a pragmatic wager: that collaboration with AI, if structured fairly, is a more viable path than resistance — and that quality journalism still commands a price worth paying.
Two of Brazil's largest news organizations — Folha de S.Paulo and UOL — have signed the country's first formal licensing agreement with OpenAI, granting the company rights to use their journalism to train and improve ChatGPT. The deal, finalized in May 2026, marks a turning point in how Brazilian media is navigating artificial intelligence and the contested question of who profits from journalistic content in the age of large language models.
For years, AI companies scraped reporting from professional newsrooms without permission or payment. Publishers had little leverage until they began organizing, and over the past two years OpenAI has negotiated licensing arrangements with major outlets across the United States and Europe. Brazil's media landscape was slower to formalize these arrangements — until now. Folha de S.Paulo, founded in 1921 and the country's largest newspaper by circulation, and UOL, a digital platform reaching millions daily, together represent a substantial share of Portuguese-language journalism flowing into global AI training data.
The specific financial terms remain undisclosed, but the structure follows OpenAI's established model: legal access to archives and ongoing content in exchange for compensation and some control over attribution within ChatGPT's outputs. For OpenAI, the deal strengthens its Portuguese-language capabilities. For the publishers, it opens a new revenue stream as traditional advertising continues to erode.
The agreement's significance extends well beyond Brazil. Publishers across Latin America will now face pressure to negotiate their own terms or risk being sidelined. It also raises harder questions about smaller outlets and independent journalists who lack the institutional weight to command favorable deals. Whether tiered licensing models will emerge to accommodate them remains unanswered.
What the deal ultimately reflects is a broader, still-unresolved negotiation between technology and the press — one in which some publishers are beginning to see AI not as a threat to be fought, but as a partner to be priced. Whether that optimism holds depends on whether the terms prove genuinely meaningful as the technology continues to evolve.
Two of Brazil's largest news organizations—Folha de S.Paulo and UOL—have become the first major publishers in the country to strike a formal licensing agreement with OpenAI, granting the company rights to use their journalism to train and improve ChatGPT. The deal, signed in May 2026, represents a watershed moment for how Brazilian media is approaching artificial intelligence and the question of who owns and profits from journalistic content in the age of large language models.
For years, news organizations worldwide have grappled with the reality that AI companies were scraping their reporting without permission or compensation. OpenAI's ChatGPT, trained on vast swaths of internet text, inevitably absorbed millions of articles from professional newsrooms. Publishers had little recourse and less leverage—until they began to organize. Over the past two years, OpenAI has negotiated licensing deals with major outlets in the United States, Europe, and elsewhere, acknowledging that quality journalism has value and that sustainable partnerships serve everyone better than uncompensated extraction.
Brazil's media landscape has been slower to formalize these arrangements, but Folha and UOL's agreement changes that calculus. Both outlets are heavyweight institutions in Brazilian journalism. Folha de S.Paulo, founded in 1921, is the country's largest newspaper by circulation and influence. UOL, a digital-native news and services platform, reaches millions of Brazilians daily. Together, they represent a significant portion of Portuguese-language journalism that feeds into global AI training datasets. Their decision to license content rather than fight or ignore the issue signals a pragmatic acceptance that AI integration into media is inevitable—and that publishers may as well be compensated for it.
The terms of the agreement have not been publicly disclosed in detail, but the structure mirrors deals OpenAI has struck elsewhere: the company gains legal rights to use the publishers' archives and ongoing content for training purposes, while Folha and UOL receive financial compensation and, presumably, some degree of control over how their work is attributed or presented within ChatGPT's outputs. For OpenAI, the arrangement strengthens the quality and diversity of its training data, particularly for Portuguese-language understanding. For the publishers, it opens a new revenue stream at a time when traditional advertising models continue to erode.
What makes this deal significant beyond Brazil is its ripple effect across Latin America. Other major publishers in the region—in Mexico, Argentina, Colombia—will now face pressure to negotiate their own terms or risk being left behind. The precedent established by Folha and UOL suggests that licensing is becoming the default path forward, not the exception. It also raises questions about smaller outlets and independent journalists who lack the bargaining power of established institutions. Will they be left out of these arrangements entirely, or will OpenAI develop tiered licensing models that accommodate different scales of operation?
The agreement also reflects a broader shift in how tech companies and media organizations are learning to coexist. Rather than viewing AI as an existential threat to journalism, some publishers are beginning to see it as a tool they can partner with—one that, if structured fairly, might actually help sustain quality reporting. Whether this optimism proves warranted depends largely on whether the financial terms are genuinely meaningful and whether the partnerships remain stable as AI technology continues to evolve. For now, Folha and UOL have made their choice: they are betting that collaboration with OpenAI is better than resistance, and that their journalism is valuable enough to command a price.
The Hearth Conversation Another angle on the story
Why does it matter that these two specific outlets made this deal first in Brazil?
Because they're not small players. Folha and UOL together represent a huge share of what Brazilians read and what gets fed into global AI systems. When they move, other publishers have to think about whether they're being left behind.
What's actually in the deal? Do we know what they're being paid?
The financial terms aren't public. But the structure is becoming standard: OpenAI gets to use their content for training, they get paid, and there's probably some say in how their work appears in ChatGPT's responses. It's a licensing model, not a one-time payment.
Is this good for journalism?
That depends on whether the money is real and whether it lasts. If publishers can actually sustain themselves partly through these deals, yes. But if it's just a small revenue stream that doesn't offset what they've already lost to digital disruption, it's more of a band-aid.
What about smaller outlets that can't negotiate like Folha can?
That's the real question. They might get nothing, or they might get swept into OpenAI's training data anyway without compensation. The deal doesn't solve that problem—it might actually make it worse by establishing that only big players get a seat at the table.
Does this change how other countries will approach this?
Almost certainly. Latin America is watching. If Folha and UOL's deal works out, you'll see similar agreements across the region. If it doesn't—if the money dries up or the terms shift—publishers will be much more skeptical.