OpenAI says ChatGPT work use is spreading across 165 activities
Home
Latest reporting
News
OpenAI’s latest Signals release maps ChatGPT work messages to 165 activities across nearly 100 countries, showing a less concentrated mix over time.
OpenAI on Monday said its latest Signals release maps work-related ChatGPT use across 165 Intermediate Work Activities and nearly 100 countries, offering a broader view of how people use the chatbot for work and how that pattern is changing over time. The company said the findings are based on its own data and are intended to show message-level work use, not employment effects or productivity gains.
Evidence and context
According to OpenAI’s publication dated 26 October 2026, the analysis uses the O*NET occupational framework to classify work-related messages from individual ChatGPT plans. It excludes enterprise messages and Codex usage. OpenAI said researchers do not read individual conversations; instead, the system applies automated classification after identifying information is removed, then aggregates results, with small-group results withheld and statistical noise added for privacy.
The company said that across 74 countries observed in every quarter, the ten most common activities accounted for an average 51.1% of classified work messages in Q3 2026, down from 56.0% in Q1 2025. OpenAI characterized that shift as evidence of a less concentrated activity mix over time. These figures are message-share statistics, not measures of employment, productivity or job displacement.
OpenAI also cautioned that comparisons with employment should be made carefully. In its background note, the company said about 43% of the employment-weighted task mix for mapped U.S. occupations falls into “Other IWAs,” meaning the individually reported activity categories do not capture the full set of tasks. The figures were presented as OpenAI’s own findings and were not described as an independent audit.
The release follows earlier OpenAI posts on worker behavior and agentic AI published on 16 September 2026 and 14 October 2026, respectively. The latest report extends that line of work by focusing on how ChatGPT use is distributed across activities and geographies rather than on a single use case or industry segment.
