OpenAI Report: Cross-Occupational AI Task Share Rises to 25.9% Among Tracked Employees

OpenAI released its latest "Work at the Frontier" report on September 16, examining how employees use AI to take on tasks beyond their traditional job boundaries. The research found that some cross-occupational AI usage becomes routine over time, with such tasks accounting for a growing share of workers' AI activity.

The four-month study tracked about 6,200 professional employees from April to July 2026 and analyzed more than 1.5 million work-related ChatGPT messages. The share of cross-occupational tasks in occupation-specific AI activity rose from 13.1% in April to 25.9% in July. In a one-month matched tracking sample, 23.6% of employees revisited a cross-occupational task they had handled the previous month, compared with 8.4% for comparable employees who had not handled the same task in the prior month.

Retention rates varied by task type. Tasks such as discussing products or services with customers (54%), writing advertising or promotional copy (44%), and creating marketing materials (37%) had higher month-to-month reuse rates, while explaining financial information had a reuse rate of about 15%, below the 18.5% average for all cross-occupational tasks.

When handling tasks outside their core duties, employees tended to give AI more examples and context, and more often asked AI to check or verify work, rather than asking for step-by-step instructions. The report suggests employees may be using AI as a way to access expertise from other fields, bringing specific problems and materials to the model.

OpenAI said the findings point to a possible path for AI-driven role evolution: workers first experiment with new tasks, find AI helpful, and then gradually incorporate those tasks into their regular workflows. The report advises companies to reconsider how work itself is designed, not just what AI tools are provided.