OpenAI releases ChatGPT work activity dataset covering 165 tasks in nearly 100 countries
OpenAI released ChatGPT work activity data on Oct 26, using the O*NET framework to map occupational tasks into 165 shared intermediate work activity categories across nearly 100 countries from the first quarter of 2025 through the third quarter of 2026. The data covers work-related messages on individual ChatGPT plans and excludes enterprise messages and Codex usage. Researchers cannot read individual conversations, classification is automated after identifiable information is removed, only aggregate statistics are published, small-group results are suppressed and statistical noise is added. Across the 74 countries with data in every quarter, the ten most common activities fell from 56.0 percent of each country's classified work messages in Q1 2025 to 51.1 percent in Q3 2026, showing usage spreading from a few activities to many while countries specialize in different directions. A separate comparison found that about 43 percent of the employment-weighted task mix for mapped US occupations falls into categories not individually reported, including moving materials, operating vehicles, and maintaining health or medical records, which did not meet the threshold for consistent presence in the US sample.