ChatGPT's Work Mode Is Twice as Action-Oriented, OpenAI's First Global Data Shows

OpenAI just dropped a fresh batch of country-by-country data, and the headline is stark: when people use ChatGPT for work, they’re more than twice as likely to be building or creating something than they are outside of work. We’re talking coding, analysis, drafting documents—actual output, not just information fishing. That “from asking to doing” shift is the core signal in their new Signals dataset, which reflects how over a billion people are using the consumer versions of the tool.

The data also kills the idea that this is just a North American or European story. Peru, Uruguay, and Costa Rica saw the biggest jumps in per-capita rankings for Q2 2026, with adoption accelerating particularly fast across parts of Latin America, Africa, and Oceania. The global gap is narrowing, and it’s narrowing quickly.

Multimedia is the dark horse here. Since the release of ChatGPT Images 2.0 in April, multimedia tasks now account for 7.8% of all messages globally. That’s the fastest-growing use case by far, and in Brazil and Colombia, it already represents more than one in ten messages. It’s a clear signal that the platform is outgrowing its text-based origins.

Demographically, the “early adopter” cliché is crumbling. The share of messages from users 35 and older is up 5% year-over-year globally, with France and Czechia seeing shifts of more than 10 percentage points. That’s not a gentle trend; that’s a rapid demographic rebalancing. For policymakers and researchers watching these numbers, the message is clear: the professionalization and globalization of AI use aren’t coming—they’re already baked into the quarterly figures.