OpenAI published new economic research analyzing more than 1.5 million work-related ChatGPT messages from April through July 2026. Among roughly 6,200 workers observed throughout, previously used cross-occupation tasks rose from 13.1% of occupation-specific AI activity in April to 25.9% in July.
OpenAI's data suggests AI use outside a worker's occupation can become a recurring part of the job rather than a one-off experiment. Workers bring context such as documents or examples, write shorter prompts, and more often ask the model to check or verify, which OpenAI interprets as borrowing expertise from another field. Workers returned to a cross-occupation task from the prior month 23.6% of the time, against 8.4% for comparable workers without prior use.
The finding matters now because organizations deciding how to adopt AI may need to redesign tasks, not only grant tool access. Return rates varied widely, from 54% for discussing goods or services with customers to about 15% for explaining financial information. Jobs could broaden in practice while titles stay the same, which complicates how teams assign review and responsibility.
The study measures message patterns from ChatGPT Business users and does not establish whether output quality or productivity improved. Stronger evidence would come from longer observation windows that track error rates and outcomes for recurring cross-occupation tasks.