OpenAI just bought credibility in the world's fifth-largest internet market, and Brazilian publishers just figured out how to get paid for being scraped.
The Summary
- OpenAI inked a content deal with Grupo Folha and Grupo UOL, two of Brazil's largest media groups, to surface their journalism in ChatGPT responses with attribution
- The partnership gives OpenAI licensed access to trusted Portuguese-language news while publishers get compensation and traffic referrals
- Brazil has 165 million internet users and represents the largest Portuguese-speaking market globally, a major gap in OpenAI's training data and real-time information access
The Signal
This is the fourth major publisher deal OpenAI has signed in 2025, following agreements with News Corp, Le Monde, and Prisa Media. The pattern is clear: OpenAI is systematically licensing content from legacy media to sidestep the legal minefield of training on copyrighted material while building regional credibility for ChatGPT as a news source.
Grupo Folha publishes Folha de S.Paulo, Brazil's third-largest newspaper. Grupo UOL runs the country's leading news portal. Together, they represent mainstream Brazilian journalism in a market where misinformation spreads faster than fact-checking can keep up. OpenAI gets two things here: clean training data in Portuguese and a defensible answer when Brazilian regulators come asking about data sourcing.
"OpenAI is buying its way out of the copyright wars, one regional publisher at a time."
The deal structure matters more than the announcement admits. Publishers get:
- Direct payment for content licensing
- Attribution when ChatGPT cites their reporting
- Referral traffic back to their sites
OpenAI gets:
- Legal cover for using their archives in model training
- Real-time news access for ChatGPT responses
- A foothold in Latin America's largest AI market
Brazil is the world's ninth-largest economy and its government has been aggressive about regulating tech platforms. Getting ahead of content licensing disputes before they become legal battles is strategic. The EU's AI Act and ongoing lawsuits from The New York Times have made "we scraped it so we can use it" a liability, not a business model.
The Implication
Watch for OpenAI to announce similar deals in India, Indonesia, and Nigeria within six months. The licensing playbook is now proven, and every major non-English market with strong regional publishers is a target. For media companies, the calculation is simple: take the licensing check now or watch ChatGPT synthesize your reporting without attribution or payment. For users, it means ChatGPT's answers get more locally relevant and harder to dismiss as Anglo-centric. And for competing AI labs, it's another moat. Content licensing at this scale requires capital and credibility most startups don't have.