MikhbarMIKHBAR
Artificial Intelligence

OpenAI Releases Signals Data on Global ChatGPT Work Use

OpenAI Economic Research has released new data to help researchers explore how consumers utilize ChatGPT for professional tasks across nearly 100 countries.

OpenAI Releases Signals Data on Global ChatGPT Work Use

New Data Release for Labor Researchers

OpenAI Economic Research has launched a new data release designed to help researchers investigate the intersection of artificial intelligence and the labor market. The newly published OpenAI Signals data explores 165 work activities across nearly 100 countries, shedding light on how consumers incorporate ChatGPT into their professional routines. According to the announcement, this information provides a novel method for studying AI's broader economic role by revealing the specific tasks users bring to the platform.

The activities documented span a wide range of professional duties, from preparing instructional materials and developing marketing content to programming and analyzing business data. Researchers can utilize this quarterly data—spanning from the first quarter of 2025 through the third quarter of 2026—to observe how ChatGPT work applications vary internationally, monitor shifts over time, and correlate the metrics with other economic indicators.

Scaling AI to uncover new research leads in complex biological data — Cover
Image related to the report from OpenAI News · Source: OpenAI News

Framework and Standardization with O*NET

To ensure compatibility with existing employment statistics, the dataset draws on O*NET, a standardized framework that groups occupation-specific tasks into broader categories. This common framework allows researchers to effectively cross-reference ChatGPT utilization with other types of occupational and employment data.

The data links ChatGPT usage to 165 Intermediate Work Activities, which are commonly utilized task categories across various occupations. Interactive charting tools built into the project enable analysis of these work activities within specific countries, tracking how activity mixes shift from broad categories down to individual tasks.

Shifting Concentration and International Specialization

Analysis of the Intermediate Work Activities revealed that work-related ChatGPT usage became less concentrated in the most common activities as time progressed. Specifically, across 74 countries observed consistently in every quarter, the ten most popular activities accounted for an average of 51.1% of each country's classified work messages in the third quarter of 2026, marking a decrease from 56.0% in the first quarter of 2025.

This trend indicates that nations tended to specialize in their professional AI use over time, with different countries focusing on distinct types of work. OpenAI notes that these variations may stem from underlying differences in the types of work performed globally or varying levels of access to AI tools.

Improving ChatGPT for low-resource programming languages — listing image
Image related to the report from OpenAI News · Source: OpenAI News

Comparing ChatGPT Usage with the U.S. Economy

The dataset also facilitates direct comparisons between ChatGPT activity messages and tasks performed across jobs in the United States. Paired visualizations illustrate the divergences between how individuals deploy ChatGPT for work versus the actual task distribution within the broader U.S. economy.

The findings show that the two distributions differ significantly. Approximately 43% of the employment-weighted task mix for mapped U.S. occupations falls into other categories outside the individually reported groupings—such as operating vehicles, moving materials, and maintaining health or medical records—which did not meet the necessary threshold for consistent representation in the original U.S. ChatGPT sample.

Rosalind Academic Plan - Card
Image related to the report from OpenAI News · Source: OpenAI News

Privacy Protections and Scope of the Data

This release expands upon OpenAI Signals, an ongoing initiative to share empirical evidence regarding human interaction with artificial intelligence. The underlying data specifically covers work-related messages originating from individual ChatGPT plans, intentionally omitting enterprise messages and Codex utilization.

To safeguard user privacy, researchers do not read individual user conversations. Automated systems classify messages only after identifying information has been stripped away. Furthermore, publishers of the dataset release aggregated statistics, withhold results tied to small groups, and inject statistical noise to preserve message confidentiality. Interested parties can explore and download the complete dataset to conduct further independent investigations.

Sources

Continue chronologically

Related entity coverage