OpenAI publishes country-level view of how ChatGPT use is changing

OpenAI has published country-by-country figures describing how people use ChatGPT, offering a broader view of an AI product that has often been discussed through headline user numbers alone. The company said the data show a shift from asking questions toward completing tasks and creating work, especially in professional settings. It also reported faster growth in multimedia use, rising use among people over 35 and gains in per-capita adoption for a number of countries that began below the earliest-adopting markets. The release describes patterns in consumer accounts rather than a complete census of all business AI activity.

At work, OpenAI said users are more than twice as likely to use ChatGPT to complete a task or create something as they are outside work. The company groups editing, coding and analysis within that task-oriented behavior. Outside work, information-seeking and clarification remain the largest category. The distinction is useful because it separates a familiar chatbot interaction from a workflow in which a person expects an output that can be used immediately. It does not mean every work result is ready without review, but it suggests that the product is increasingly being placed inside everyday production rather than used only for exploration.

The country rankings point to a second change: adoption is becoming less concentrated among the markets that embraced generative AI first. OpenAI said countries in parts of Latin America, Oceania and Africa increased their per-capita standing faster during the second quarter of 2026, with Peru, Uruguay and Costa Rica among the largest movers. The company described this as a narrowing adoption gap rather than a reversal of growth in North America and Europe. That qualification is important. A rising rank measures relative movement, and it does not by itself show that every country has the same access, connectivity, language support or economic opportunity around AI.

Multimedia was the fastest-growing use case in the figures, according to OpenAI. Following the April release of ChatGPT Images 2.0, the company said multimedia generation, analysis and retrieval accounted for 7.8% of messages globally. The share was above one in ten messages in countries including Brazil and Colombia. The result gives a more concrete indication of how new product capabilities can change behavior: users appear to be moving beyond text prompts when they have tools for creating, examining or retrieving media. Still, the company said practical guidance, writing and information-seeking remained the leading use cases.

The age data add another dimension to the release. OpenAI said the share of age-classified messages from users older than 35 has risen in almost every country over the last year, reaching a five-percentage-point increase overall. In France and Czechia, it said the share rose by more than 10 percentage points. These figures rely on users who self-reported their age, so they should be understood as a measure of the eligible sample rather than a statement about every ChatGPT user. Even so, the trend challenges the assumption that generative AI is used mainly by a small, young group of early adopters.

The data are being published through OpenAI Signals, the company’s research platform, and cover messages in ChatGPT Free, Go, Plus and Pro accounts generally managed by individuals. That scope excludes a large portion of enterprise deployment, where access may be organized through employers, education systems or APIs. The release is therefore most useful as a window into consumer and individual professional behavior. It offers comparative information that policymakers, researchers and product teams can examine, while leaving open questions about how well classified message categories capture the variety of real work people perform with AI.

For the broader market, the strongest message is that adoption is changing in both geography and purpose. ChatGPT use is becoming more global, more connected to task completion and more multimedia-oriented, while a wider age range is participating. None of those shifts settles whether AI is improving work or learning in every context; the figures do not measure quality, accuracy or downstream outcomes. But they do show why AI companies are increasingly competing on integration and usability as much as on model capability. When a product moves from answering questions to helping complete work, reliability, review and clear limits become more consequential than a simple message count.