A country-level view of ChatGPT use
On August 6, OpenAI added its first country-level usage analysis to OpenAI Signals, using privacy-preserving aggregate classification. According to the company’s taxonomy, people using ChatGPT at work were more than twice as likely to make “doing” requests—creating, analyzing or completing work—than people using it outside work.
Multimedia represented 7.8% of messages globally and more than one in ten in Brazil and Colombia. OpenAI also reports that the global share of users over 35 rose five percentage points year over year, with increases above ten points in France and Czechia.
Product and workplace implications
The pattern suggests workplace adoption is moving beyond question answering into drafting, organizing data and completing repeatable tasks. Language, mobile access, creative economies and labor structures may shape feature use, so one global average is a weak basis for localization.
Boundaries of the dataset
The analysis covers Free, Go, Plus and Pro individual accounts; Enterprise and workspace activity is excluded. “Work” and “doing” depend on OpenAI’s proprietary classifiers. This is neither a population-representative survey nor independent statistics, and message shares do not prove productivity or economic impact. Organizations should measure completion time, correction rate, errors and cost in their own workflows.