Workers are using AI to tackle tasks far beyond their traditional job descriptions, according to new research from OpenAI.

The company analyzed more than 800,000 messages from US ChatGPT users and found that 43.5% of occupation-specific AI use involves tasks typically associated with other roles. The pattern suggests AI is reshaping who does what at work, not just how work gets done.

"Task crossover" appears most pronounced among customer experience workers, where 77% of occupation-specific messages involve outside tasks. Designers follow at 75%, then human resources workers at 69%.

Marketing and engineering tasks travel furthest across organizational boundaries. Financial calculation and technology troubleshooting each appear among the top three outside tasks in seven different occupation groups.

Small businesses show higher crossover rates

The research reveals stark differences by company size. Workers in organizations with 2-5 seats devote 18.9% of their AI use to outside-occupation tasks, compared to 16.3% in companies with over 100 employees.

OpenAI suggests smaller organizations lack specialist resources, making AI particularly valuable as a "generalist tool where specialist resources are scarce."

The study excludes generic activities like writing and scheduling that appear across many roles. Among the remaining occupation-specific messages, nearly half cross traditional job boundaries.

Design workers illustrate one crossover pattern — they draw heavily on outside tasks but design work rarely appears elsewhere. Engineering shows the reverse: only 18.5% of engineering messages involve other fields, but engineering tasks account for 7.4% of messages from other occupations.

Marketing stands out in both directions, with workers combining tasks from multiple domains while marketing work spreads widely across organizations.

The findings come from OpenAI's new "Work at the Frontier" research series, which uses ChatGPT usage data to track real-time changes in work patterns. The company plans regular reports based on this unique dataset.

OpenAI frames the research as evidence that many jobs will "reorganize" rather than disappear, with day-to-day tasks changing substantially as AI adoption accelerates.