OpenAI signs a strategic media deal in Brazil

OpenAI expands its media diplomacy in Latin America with a landmark deal in Brazil

OpenAI has announced a new content partnership with Grupo Folha and Grupo UOL, two major players in Brazil’s media landscape. The news was made official by OpenAI in a communication dedicated to this alliance, presented as a strategic agreement aimed at integrating reliable journalistic content into ChatGPT, with explicit source attribution. Behind this announcement, this is not just another licensing contract. The move is part of a broader strategy to normalize relations between generative AI players and the press, at a time when the question of access to content, its compensation, and its visibility is becoming central.

The choice of Brazil is far from incidental. With more than 200 million inhabitants, strong digital penetration, a large online advertising market, and a media ecosystem structured around major national brands, the country represents one of the most strategic grounds in Latin America for technology platforms. It also constitutes a relevant laboratory for AI use in Portuguese, a segment that is still less saturated than English but large enough to weigh in OpenAI’s commercial decisions.

The two selected partners embody this centrality precisely. Grupo Folha notably publishes Folha de S.Paulo, one of the country’s most influential newspapers, founded in 1921 and often regarded as a benchmark in Brazil’s general-interest press. Grupo UOL, for its part, operates UOL, a long-standing information and digital services portal that has been deeply embedded in the country’s web usage since the 1990s. By bringing together a reference news outlet and a digital player with a very large audience, OpenAI secures both editorial credibility and volume.

The wording of the announcement is also revealing. OpenAI emphasizes the integration of reliable journalistic content into its products, with mention of sources. This vocabulary responds to several criticisms directed at conversational assistants since the explosion of generative AI at the end of 2022: a tendency to produce unsourced answers, difficulty distinguishing between synthesis and quotation, and publishers’ concerns about platforms capturing value. In other words, the Brazilian deal should not be read only as an enhancement of the ChatGPT user experience, but as another piece in a policy of legitimization.

This policy is now international. After several agreements concluded in English-speaking markets, OpenAI is visibly accelerating beyond its historical core. Brazil appears here as an important milestone: a large non-English-speaking market, with high media intensity, where the question of trust in information is particularly sensitive. In this context, media partnerships are becoming a lever that is at once business, regulatory, and reputational.

What the agreement with Grupo Folha and Grupo UOL provides for

According to OpenAI’s announcement, the partnership is intended to allow ChatGPT to display and use content from Folha de S.Paulo, UOL, and other brands associated with the partner groups, with links and clear attribution to the original sources. The stated objective is twofold: improve the informational quality of the answers provided to users and guarantee direct visibility for the publishers whose content feeds those answers.

The most important point is probably the content licensing logic. Since 2023, major publishers have been demanding more explicit contractual mechanisms to govern the use of their productions by AI systems. Generative models have been trained on massive volumes of text from the web, which has triggered a series of disputes, copyright debates, and private negotiations. By highlighting a formal agreement with two Brazilian groups, OpenAI is seeking to show that the integration of press articles into its products can take place within a negotiated, identifiable, and potentially reproducible framework.

OpenAI has not publicly detailed all the financial terms of the operation, which is common in this type of agreement. The amounts, the exact duration of the license, the indexing conditions, or any exclusivities were not made public in the initial communication. However, the strategic message is clear: the company wants to offer answers enriched by licensed press content, rather than relying only on freely accessible information, heterogeneous sources, or older data. This makes it possible both to strengthen the timeliness of answers and to reduce, at least in part, criticism over the unpaid exploitation of journalistic content.

For the Brazilian groups, the interest is also tangible. Publishers have been seeking for several years to rebalance their relationship with digital platforms. After long depending on traffic from search engines and social networks, they are seeing a new layer of intermediation emerge: AI assistants. The risk is well known: if the user gets a synthesis directly in the ChatGPT interface, they may no longer click through to the original site. Explicit attribution, the presence of links, and recognition of the media brand therefore become essential to preserve audience, editorial value, and bargaining power alike.

In its communication, OpenAI insists on the idea of helping users discover and explore the news through more reliable answers. This wording is important, because it seeks to position ChatGPT not as an absolute substitute for media, but as an access and guidance interface. This is a line of defense increasingly used by AI companies: they would not replace the press, they would facilitate access to it. In practice, however, everything will depend on product implementation, the level of detail in the answers, and the volume of traffic actually sent back to publishers.

The partnership also comes in a context where OpenAI is gradually deploying more advanced search functions. As ChatGPT moves closer to an answer engine capable of navigating the news, the quality and legitimacy of sources become decisive. Striking deals with recognized groups helps secure this ramp-up. In Brazil, that means having content from established players, with strong national recognition and the capacity to cover politics, business, sports, and culture at scale.

• Partners involved: Grupo Folha and Grupo UOL
• Planned use: integration of journalistic content into ChatGPT
• Key promise: clear source attribution and links to original content
• Main issue: strengthen informational reliability and formalize licensed access
• Strategic scope: extension of OpenAI’s media-deal policy beyond English-speaking markets

A turning point in the relationship between generative AI and the press, after months of tension

To measure the significance of this Brazilian agreement, we need to go back over the sequence of the last 18 to 24 months. Since the meteoric rise of ChatGPT starting in November 2022, press publishers have adopted contrasting positions toward generative models. Some quickly blocked access to their sites for the crawling bots of certain AI companies. Others entered licensing discussions. Still others chose the legal route, considering that the past use of their content for training or answer generation constituted a violation of their rights.

The most emblematic case is probably that of the New York Times, which sued OpenAI and Microsoft at the end of 2023. This complaint marked a symbolic turning point: it showed that the conflict between generative AI and the press was no longer a theoretical debate, but a concrete confrontation over intellectual property, competition, and the economic value of information. At the same time, other groups chose negotiation. OpenAI thus multiplied agreements with several publishers and press groups, seeking to build a network of partners rather than let a generalized face-off with the media industry take hold.

The agreement with Grupo Folha and Grupo UOL fits precisely into this second logic. It represents a pragmatic response to a criticism that has become structural: can an AI assistant claim to help users stay informed if it does not rely on identified, authorized, and attributed content? The debate is not only legal. It is also editorial. Generative models produce fluid, sometimes convincing answers, but their relationship to the source has long been opaque. For current-events topics, that opacity is particularly problematic.

The use of licensed content makes it possible to address several issues at the same time. First, the freshness of information. An agreement with publishers gives access to recent, updated, contextualized content. Next, traceability. The user can see where the information comes from, which is essential in an environment saturated with disinformation and synthetic content. Finally, responsibility. By relying on recognized press brands, OpenAI is trying to show that it no longer merely aggregates the web indiscriminately.

This dynamic is not unique to OpenAI, moreover. The entire sector is moving in the same direction, with nuances. Google has long developed partnership programs with publishers, well before the era of generative AI, through Google News, Google News Showcase, or various support funds. With the arrival of AI Overviews and its Gemini tools, the Mountain View firm now faces similar questions about source visibility and traffic impact. Microsoft, through Bing and its AI integrations, faces the same trade-offs. Perplexity, for its part, has also sought to conclude agreements with media outlets, notably in response to criticism over the reuse of journalistic content in its search answers.

The difference, in OpenAI’s case, lies in the symbolic weight of ChatGPT. With hundreds of millions of users claimed worldwide according to the company’s recent communications, the service has become for part of the public a gateway to information, research, and decision support. This gives media agreements an importance that goes far beyond simple product functionality. They contribute to defining what the dominant informational interface of the decade will be: a web of links, a search engine, or a conversational assistant fed by contracted sources.

By naturally citing its own announcement on the partnership with Grupo Folha and Grupo UOL, OpenAI highlights a simple idea: the quality of ChatGPT’s answers on current events depends on access to reliable journalistic content that is clearly attributed and comes from formal agreements.

Brazil adds an additional dimension to this evolution. Unlike the United States or the United Kingdom, where agreements between platforms and publishers are already heavily publicized, the Brazilian market makes it possible to test the internationalization of this model in a non-English-speaking environment, with its linguistic, regulatory, and economic specificities. For OpenAI, it is a way of demonstrating that its strategy is not reserved for major North American titles.

Why Brazil is a key market for OpenAI and for publishers

Brazil combines several characteristics that make it particularly attractive ground for OpenAI. First, its demographic size: with around 203 to 215 million inhabitants depending on the references and statistical periods used, the country is by far the largest Portuguese-speaking market in the world. Next, its digital culture: mobile usage is massive, social networks play a central role, and major news portals retain notable influence. Finally, its media economy is dominated by a few powerful players capable of negotiating at the national level.

Folha de S.Paulo is not just any partner. The newspaper, created in 1921, has lived through the major political and economic cycles of modern Brazil. Its weight in public debate gives it strong symbolic value. UOL, launched in 1996, is one of the pioneers of the Brazilian internet. Its portal combines news, services, sports, entertainment, and digital products, with considerable distribution capacity. For OpenAI, the combination of these two groups makes it possible to cover both editorial authority and audience scale.

The Brazilian market also presents specific challenges in terms of disinformation and political polarization. In recent years, the country has been one of the most closely watched theaters for the viral circulation of false or misleading content, particularly during major electoral deadlines. In such an environment, a conversational assistant that claims to help understand the news cannot afford to remain vague about its sources. The agreement with recognized journalistic groups therefore also constitutes a message addressed to authorities, advertisers, and the public: OpenAI wants to be perceived as a more responsible relay of information.

There is also a logic of linguistic diversification. The generative AI industry has long been dominated by English, both in datasets and in publicized use cases. Yet future growth will largely play out in other languages. Portuguese is spoken by more than 260 million people worldwide, including of course Brazil but also Portugal, Angola, Mozambique, and other Portuguese-speaking countries. Strengthening the quality of ChatGPT’s answers in Portuguese, particularly on current events, is therefore an investment that goes beyond the Brazilian market alone.

For local publishers, the agreement can be read as an attempt to avoid the scenario already seen with social networks: depending on global intermediaries that capture attention without sharing enough value. Brazilian media, like their European counterparts, know that conversational uses will grow. If they want to retain a visible place in this new distribution chain, they must negotiate early. An agreement today can serve as leverage to influence tomorrow’s standards: attribution, links, compensation, excerpt granularity, access to audience data, and possible restrictions on certain uses.

This logic is particularly important for French-speaking and European markets, which are closely watching how these agreements are being structured outside the Anglo-Saxon world. Brazil shows that OpenAI is no longer content with signing deals only with highly visible global brands in the United States. The company is now seeking to build an international network of trusted partners capable of feeding ChatGPT in several languages and across several major news basins.

A competitive battle where content licensing becomes a strategic advantage

The partnership announced in Brazil must also be read through the prism of competition. The generative AI market is no longer being fought only on model power or token cost. It is increasingly being fought on access to premium content that is up to date and legally secured. In this new phase, media agreements are becoming strategic assets comparable, in some respects, to distribution agreements or exclusives in other digital industries.

OpenAI has an interest in accumulating these partnerships for several reasons. First, they improve the perceived quality of ChatGPT on current-events queries, an area where hallucinations or inaccuracies are particularly damaging. Next, they strengthen the company’s credibility with regulators, who are increasingly scrutinizing the conditions of access to data and the competitive effects of large models. Finally, they create a form of product differentiation: an assistant that relies on recognized sources, with attribution and links, may appear more reliable than a less transparent competitor.

This battle is unfolding in parallel with the offensive of already established giants. Google has a historical advantage in web indexing and publisher relations, but it must now adapt that legacy to a generative answer interface. Microsoft, by integrating AI into Bing and its productivity products, is betting on another form of distribution. Meta, even if its approach is different and more centered on open models and social platforms, remains a player capable of redefining access to information through its interfaces. Perplexity, smaller but very aggressive, has popularized a conversational search experience heavily centered on sources, which has put pressure on other players to make their citations more visible.

In this context, OpenAI’s Brazilian agreement serves several functions. It feeds the product, improves the image, and strengthens a commercial argument. For users, the promise is simple: more reliable answers on Brazilian current events, with identifiable references. For publishers, the proposition is more complex: monetize access to their content and preserve their visibility in a world where the conversational interface risks reducing the traditional click. For OpenAI, the challenge is to make these interests converge without slowing adoption.

It should also be noted that the race for licensed content could create new asymmetries. Large press groups, capable of negotiating directly with platforms, will more easily obtain agreements, visibility guarantees, and possibly revenue. Smaller, specialized, or regional media risk being left aside, unless more open mechanisms emerge. This question is particularly sensitive in Europe, where media pluralism is the subject of strong political attention. If conversational AI concentrates exposure on a few major partner brands, it could reinforce power gaps within the information ecosystem.

Another key element: measuring the real value of these agreements remains unclear. Publishers need to know whether attribution in ChatGPT actually generates traffic, subscriptions, or at least a measurable increase in awareness. Without precise indicators, the risk is signing agreements that legitimize the use of content without truly compensating for the potential loss of direct visits. This is one of the points the entire sector will be watching in the coming months, in Brazil as elsewhere.

The implications for France, Europe, and French-speaking markets

Seen from France and more broadly from French-speaking Europe, the agreement between OpenAI, Grupo Folha, and Grupo UOL has significance that goes beyond the Brazilian market alone. It confirms that media agreements linked to generative AI are entering a phase of international industrialization. Yet Europe already has a particular experience of power struggles between platforms and publishers, notably through the debate over press neighboring rights.

In France, this issue took on a strong political and regulatory dimension as early as the end of the 2010s. Negotiations between Google and French publishers around neighboring rights showed that a legal framework can force major platforms to recognize part of the value produced by the press. The arrival of conversational assistants updates this debate in another form. It is no longer only about displaying excerpts in a search engine, but about the synthesis, reformulation, and redistribution of information in a conversational interface.

For French, Belgian, Swiss, or Quebec press groups, the Brazilian example sends several signals. First, OpenAI is ready to sign beyond English-speaking markets. Next, the company seems to want to make source attribution a central element of its communication. Finally, the selection of powerful national partners suggests that negotiations will first favor players capable of bringing together notoriety, content volume, and editorial legitimacy.

The French-speaking market nevertheless has specific features. It is more fragmented linguistically and geographically than a national market like Brazil. France remains the main center of gravity, but the French-speaking world also includes Belgium, French-speaking Switzerland, Luxembourg, French-speaking Canada, and many African countries. For OpenAI, building a solid French-language news offering therefore requires either concluding several complementary agreements or relying on a few major groups with transnational reach.

On the regulatory front, the European Union adds an extra layer of complexity. Between the Digital Services Act, the Digital Markets Act, copyright rules, and now the AI Act, AI companies must demonstrate more transparency, diligence, and responsibility. In this framework, licensing agreements with publishers can play a stabilizing role. They make it possible to show that access to content does not rely only on a unilateral logic of scraping or aggregation, but also on explicit contractual relationships.

For European media, there is an opportunity here but also a risk. The opportunity is to turn a technological constraint into a new source of revenue and visibility. The risk is entering into renewed dependence on interfaces controlled by a few American companies. The recent history of digital has shown that platforms regularly change their distribution rules, sometimes abruptly. Publishers that sign with AI players will therefore want to obtain stronger guarantees than those that prevailed in the era of social distribution.

The Brazilian case may also interest European groups on another level: language. If OpenAI is investing in agreements in Portuguese, that means informational quality in non-English languages is becoming an axis of competition. For French, Spanish, German, or Italian, the same logic should follow. Publishers with rich archives, recognized newsrooms, and strong brands could find themselves in a position of strength to negotiate. Conversely, those that delay risk seeing standards set without them.

Beyond the announcement, what this strategy reveals about the future of AI assistants

The agreement between OpenAI, Grupo Folha, and Grupo UOL is not just one more corporate news item in the long list of partnerships signed by AI giants. It sheds light on a deeper transformation: the conversational assistant is becoming an information access infrastructure. And when a product occupies that place, it can no longer be content with being technically high-performing; it must also be economically, legally, and socially acceptable.

OpenAI seems to have understood that the next phase of ChatGPT’s growth will depend as much on trust as on performance. The trust of users, who want accurate, sourced answers. The trust of publishers, who want to be recognized and compensated. The trust of regulators, who want to avoid a new opaque concentration of access to information. The Brazilian partnership responds to all three dimensions at once, even if it does not fully resolve them.

In the long term, several scenarios are taking shape. The first would be a generalization of bilateral agreements between major models and major publishers, resulting in a global map of licensed content. In that case, AI assistants would become hybrid distributors, halfway between the search engine, the news aggregator, and the personalized digital newsstand. The second scenario would be more conflictual: multiplication of disputes, fragmentation of access to content, and the rise of closed ecosystems where certain information would be accessible only through exclusive agreements. A third scenario, more likely, would see private contracts, regulatory obligations, and new technical standards for citation, opt-in, and opt-out coexist.

In all cases, the value of the source is becoming central again. For years, the web accustomed users to the rapid circulation of information, often detached from its production context. Generative AI pushes this movement even further by turning articles into synthetic answers. Partnerships like the one signed in Brazil indicate that the industry is now trying to reintroduce visibility, hierarchy, and contractualization into this chain. That does not mean the problem is solved. It means the battle is shifting: from raw access to content toward the conditions of its integration into intelligent interfaces.

For OpenAI, Brazil could serve as a reproducible model in other major non-English-speaking markets. The company has an interest in demonstrating that its approach works as well in Portuguese as in English, as well in an emerging market as in a mature Western economy. If the indicators of usage, satisfaction, and publisher relations are judged positive, we can expect an acceleration of this type of agreement in continental Europe, Asia, and Latin America.

One decisive question remains: who will control the final relationship with the user? Media produce the informational raw material, but the assistant controls the interface, personalization, the context of the query, and ultimately part of the attention. That is where the real economic balance of the next decade will be decided. By signing with Grupo Folha and Grupo UOL, OpenAI is not just obtaining quality Brazilian content; the company is consolidating its position in a global race where the conversational interface seeks to become the natural entry point to the news. For French-speaking publishers as for European regulators, the signal is clear: negotiation over the place of the press in AI systems is only just beginning, and it will now play out internationally, language by language, market by market, against the backdrop of a lasting redefinition of value sharing between information producers and intelligent platforms.