On July 27, 2026, OpenAI released the first installment of a new research series called "Work at the Frontier," which analyzes how AI is changing the way people work[1]. Examining more than 800,000 messages from U.S. ChatGPT users, the study found that 43.5% of occupation-specific messages concerned work traditionally handled by a different profession. Just as a salesperson steps into data analysis or a marketer fixes a website glitch on their own, AI is dissolving the very boundary of who does which job.

What Is "Task Crossover" Across Occupations

OpenAI's economic research team analyzed more than 800,000 messages from U.S. ChatGPT users[1]. It found that 16.8% of all work-related messages, and 43.5% of occupation-specific messages, concerned tasks tied to an occupation other than the user's own.

The company calls this phenomenon "task crossover": work once associated with one occupation appearing in the AI use of people in another. In the analysis, broadly shared activities such as writing, summarizing, and scheduling were first separated out as "generic." For the remaining occupation-specific messages, the researchers judged whether each fell inside or outside the user's own occupation. The share falling outside reached 43.5%[1].

OpenAI frames the study as evidence that, unlike much research that fixes a list of tasks for a given occupation and asks whether AI can perform them, the question of who takes on which task is itself changing.

Pronounced "Borrowed Work" in Customer Support, Design, and HR

The share of outside-occupation work runs especially high in several roles[1]. After excluding generic activities, the proportion of occupation-specific messages devoted to outside-occupation tasks was as follows.

Among customer experience workers it was 77%, among designers 75%, among human resources staff 69%, among legal workers 56%, and among marketers 53% — all involving work tied to another profession. These occupations are effectively "borrowing" tasks from other roles through AI. Work that once required a handoff to a specialist is now done by the person who first encounters the need, and that shift in the division of labor is showing up in AI usage.

Zooming in on specific tasks, financial calculation and technology troubleshooting each ranked among the top three outside tasks in all seven other occupation groups analyzed. Marketing work also traveled widely: creating marketing materials appeared across five other groups and was especially prominent among designers[1].

Occupations That "Import" Work and Those That "Supply" It

Viewed as larger bundles of tasks, two directions emerge[1]: occupations that bring in many tasks from other fields, and occupations whose own work spreads across many jobs.

Design illustrates the first pattern. About 35.2% of designers' messages involve work usually tied to another occupation, while design tasks account for just 1.7% of messages from workers in other fields. Designers draw heavily on outside tasks, but design work itself rarely appears elsewhere.

Engineering is nearly the reverse. Only 18.5% of engineers' messages involve tasks from other fields, but engineering tasks make up 7.4% of messages among workers in other occupations. From troubleshooting software to handling technical systems, engineering is an important source of work that others take on.

Marketing stands out in both directions. Marketers devote 24.3% of their messages to tasks associated with other occupations, while marketing tasks account for 8.9% of messages among workers in other fields — the highest outward share in the sample[1].

Company Size Shapes How AI Is Used

The size and structure of a business also affect how AI changes work[1]. Large companies offer specialized teams, established workflows, and internal services, while in a smaller organization the person closest to a problem is more likely to take it on themselves.

Among average users, the outside-occupation share fell from 18.9% in workspaces with 2–5 seats to 16.3% in workspaces with more than 100 seats. Among the heaviest users, however, the same one-directional pattern did not appear. OpenAI notes that AI may be especially useful as a generalist tool where specialist resources are scarce[1].

An Early Signal of Change

OpenAI describes such usage data as a leading indicator of where work is shifting[1]. Before firms rewrite job descriptions or create new titles, workers are already experimenting with new combinations of tasks. The company reads this as AI usage patterns revealing occupational change earlier than conventional labor-market statistics will capture it.

Outside observers have weighed in as well. According to the U.S. outlet Axios, OpenAI chief economist Ronnie Chatterji said the boundaries between jobs are likely already becoming more flexible because of AI[2]. That said, the data reflects what users asked ChatGPT to do, not whether those tasks were completed to a high standard. The spread of boundary-crossing use is better read as a widening of experimentation — people trying on new roles — than as proof of results.

Summary

OpenAI's new study shows that 43.5% of occupation-specific ChatGPT use reaches into work traditionally handled by other professions, with the trend strongest in customer support, design, and HR. AI is changing not just how work gets done but who does it, and that change surfaces in usage data as an early signal before job descriptions are formally rewritten. How far the ongoing "Work at the Frontier" series can keep charting this structural shift will be worth watching.

Source: https://openai.com/index/how-ai-is-expanding-what-people-do-at-work

Source: https://www.axios.com/2026/07/27/openai-chatgpt-work-specialists