OpenAI Signals Data Shows ChatGPT Use Deepening Worldwide

Six months after signing up for ChatGPT, users send 50% more messages per day than they did in their first month, according to new OpenAI Signals data. That figure is the clearest evidence yet that AI adoption is not just widening but deepening: people who stick with ChatGPT do more with it over time, not less.

The data, drawn from aggregated usage of Individual ChatGPT plans — Free, Go, Plus, and Pro — shows that users also doubled the number of distinct tasks they tried within six months of creating an account. OpenAI measures task breadth using a classifier that sorts messages into one of 53 capability categories.

The release matters beyond OpenAI's own metrics. Regulators, economists, and rival AI labs have lacked reliable longitudinal data on how ordinary people actually use chatbots at global scale. OpenAI positions Signals as a response: an "ongoing effort to ensure that researchers and policymakers have the best data at their fingertips to understand how AI is affecting and will impact the economy."

Adoption accelerated fastest in Africa and Asia

Every continent has seen weekly active user growth since July 2023, the baseline month OpenAI uses for its regional comparisons. In relative terms, Africa and Asia grew fastest.

The same pattern holds when countries are grouped by the UN's 2023 Human Development Index. Lower-HDI countries posted the fastest relative growth in weekly active users since July 2023. OpenAI ties this partly to pricing: the company says it "has continued to provide low-cost access to ChatGPT through our free and Go plans." For AI distribution debates — including concerns that frontier AI benefits wealthy countries first — the data suggests the opposite dynamic is taking hold, at least in relative growth rates.

OpenAI counts a user as active if they send a message in the seven days before the start of each month. Countries where ChatGPT does not operate are excluded from the analysis, as are users under 18.

English is no longer the majority language on ChatGPT

The user base has crossed a symbolic threshold. Users predominantly using a language other than English now represent over half of active ChatGPT users. Spanish, Portuguese, and Arabic lead among non-English languages.

The fastest-growing languages by share of active users since July 2023 are Uzbek, Kazakh, and Burmese — languages largely absent from mainstream AI coverage, and a signal that ChatGPT usage is penetrating well beyond the early-adopter, English-speaking demographic. That calculation is limited to languages with at least 1 million users active in June 2026, with each language's share indexed to its July 2023 level. A user's primary language is determined by the language used in the plurality of their messages.

For OpenAI, the shift carries commercial weight: the company's growth story increasingly depends on non-English markets, which shapes its product roadmap on translation quality, local pricing, and regional availability.

Gender balance, estimated through names

OpenAI also published an unusual gender estimate. Because the company does not collect information on users' gender, it inferred usage patterns from name-to-gender crosswalks, counting users with typically feminine or masculine names.

By that measure, people with typically female names now represent most ChatGPT usage globally. Brazil, Colombia, Poland, and Namibia rank among the countries where messages from users with typically feminine names most exceed those from typically masculine names. Pakistan, Bangladesh, Angola, the Democratic Republic of Congo, and Mali show the most concentrated usage by typically masculine names.

The method has clear limits. The analysis excludes users whose names are not classifiable as typically masculine or feminine, and omits countries where at least half of active users' names cannot be classified. OpenAI itself frames the numbers as "our best estimate." Still, the finding matters for a persistent question in AI research: whether chatbot usage skews heavily male, as early internet-era technology adoption often did. By OpenAI's estimate, it no longer does.

Methodology, in brief

The depth-and-breadth figures rest on a 0.1% sample of users whose accounts were created between October 15, 2025 and May 1, 2026, with activity tracked through May 31, 2026. Depth measures messages sent in the previous 28 days relative to the first 28 days after account creation. Breadth measures cumulative capabilities used, relative to those used in the first 28 days. Banned users, users under 18, and users who send no messages in their first 28 days are excluded.

All datasets are available for download, and OpenAI is inviting researchers to use them for their own analysis, with a subscription option for future data releases.

The stakes run in two directions. If OpenAI's numbers hold up under independent scrutiny, they strengthen the case that conversational AI has become routine infrastructure for hundreds of millions of people — and that its center of gravity is shifting outside the English-speaking world. Whether that translates into measurable economic effects, rather than engagement metrics, is the question Signals' next releases will have to answer.