43.5% of job-specific ChatGPT work crosses occupational lines

The sharpest signal in OpenAI’s new workplace research is not a productivity claim. It is a boundary shift. In a July 27, 2026 report, OpenAI Economic Research analyzed more than 800,000 messages from U.S. ChatGPT users and found that 16.8% of work-related messages, and 43.5% of occupation-specific messages, involved tasks associated with a different occupation.

The report, Work at the Frontier: How AI is Expanding What People Do at Work, calls this pattern task crossover. The researchers first filtered out generic work such as writing, summarizing, and scheduling, then asked whether the remaining task belonged inside or outside the user’s own occupation. That distinction matters because it captures a change that may appear before firms rewrite job descriptions or create new roles.

The strongest crossover showed up in roles that often sit close to business problems but depend on other functions for execution. Once generic tasks were excluded, outside-occupation work accounted for 77% of occupation-specific messages from customer experience workers, 75% from designers, 69% from human resources workers, 56% from legal workers, and 53% from marketers. In practical terms, a salesperson can inspect customer data, a marketer can troubleshoot a site, and a small-business owner can draft copy, review a contract, or run basic financial analysis without waiting for a specialist handoff.

The flow is not symmetrical. Designers borrow heavily from other roles, while design tasks rarely show up elsewhere. Engineering is closer to the reverse: only 18.5% of engineering messages involved tasks from other fields, but engineering tasks represented 7.4% of messages among workers outside engineering. Marketing stood out in both directions, with 24.3% of marketers’ messages involving other occupations and marketing tasks making up 8.9% of messages from other fields.

Company size adds another clue. Among average users, outside-occupation task share fell from 18.9% in workspaces with 2–5 seats to 16.3% in workspaces with more than 100 seats. The report frames AI as a generalist tool where specialist resources are scarce. The next question is whether managers treat this as informal improvisation or redesign workflows around it.