OpenAI opens a formal Brazil operation: what surging usage means for small business

Brazilians send 215 million messages a day to ChatGPT, and OpenAI just announced a formal business operation in the country. Rising usage without a defined process is where most small businesses stand today.

August 21, 2026|8 min read

What OpenAI announced

On Monday, August 17, 2026, OpenAI formally announced a business operation in Brazil, with local hiring, including implementation engineers dedicated to helping companies adopt its AI tools, along with partnerships already in place with Nubank, the IMPA research institute, and Hospital das Clínicas at USP. According to OpenAI itself, Brazilians now send around 215 million messages a day to ChatGPT, up roughly 54% from the 140 million daily messages recorded in November 2025.

The growth isn't just individual users. OpenAI's corporate base in Brazil grew fivefold over the past year, driven by adoption of AI agents inside companies. Brazil is also now the second-largest developer market for OpenAI's API in the world, ranks among the top three markets by user count (behind only the United States and India), leads global usage of ChatGPT Images, and is the largest Latin American market for Codex, the company's coding tool. Portuguese is already the third most-used language on ChatGPT globally.

One caveat worth noting: these figures, "fivefold corporate growth," "usage nearly doubling," come from OpenAI itself, not from an independent third-party audit. That doesn't make them false, but they deserve to be read for what they are: a company telling its own growth story.

Why this matters for small business, not just developers

The temptation is to read this as a story for developers or tech companies. It isn't. A formal business operation in Brazil, with a local team and institutional partnerships, typically means support in Portuguese, more accessible pricing for mid-sized and small companies, and a product built for local use cases, not just a translated version of an American product.

The most revealing data point isn't the message volume, it's the fivefold corporate growth. That points to companies, including small and mid-sized ones, already using generative AI inside their operations, often without a formal process, without a defined usage policy, without a clear picture of exactly what data is being shared with which tool. Usage growing faster than governance is exactly the scenario where problems show up: sensitive information leaking out, dependence on a tool without a proper contract, business decisions made on an AI response without verification.

What "using AI" actually looks like in a Brazilian small business today

In practice, most small businesses using AI today do it in a fragmented way: one employee uses ChatGPT to draft emails, another uses it to summarize a contract, a third uses a different tool to generate an image for a social post. Each individual use looks small and harmless. Added up, this is real AI adoption happening without the business owner ever formally deciding to adopt AI, and without any process in place to decide what can and can't be typed into a tool like that.

With OpenAI formalizing its commercial presence in the country, the trend is for this informal adoption to turn into a direct commercial offer: business plans, dedicated support, integration with existing systems. That's a good thing: it lowers the friction for a company that wants to adopt AI in a structured way. But it also raises the pressure to decide quickly, and a quick decision made without process tends to repeat the same problem at a larger scale.

Three questions before expanding AI use at your company

First: who at the company is already using a generative AI tool today, for what, and with what data? Mapping this out, even informally, just by asking the team directly, already reveals the real size of the exposure, which is usually bigger than the owner imagines. Second: is any sensitive information (customer data, contracts, financial information) being typed into an AI tool without a business contract guaranteeing that data won't be used to train a model? Third: if the company formalized AI use today, with a paid plan, a usage policy, and team training, would that change how the team works, or would it just formalize what's already happening informally?

Unstructured AI adoption costs more than it looks like

The risk of adopting generative AI in a small business usually isn't catastrophic, it's not a hack, not a breach of millions of records. It's quieter: customer information shared unnecessarily, a business decision made on a response that sounds right but was never verified, dependence on a tool that changes its price or policy without warning (as recently happened with WhatsApp Business's own AI agent). Formalizing usage, choosing a tool with a clear contract, defining what can and can't be shared, training the team, costs time upfront. Not formalizing it costs exposure later, in a way that only shows up once it's already caused a problem.

Where Diglion comes in

Diglion helps small and mid-sized businesses structure generative AI adoption safely: mapping where a tool is already being used informally, defining a simple and enforceable usage policy, and choosing the right integration between AI tools and the systems the company already runs, without stalling the productivity that motivated the adoption in the first place.