OpenAI‘s latest Work at the Frontier research examines whether workers who use AI for tasks outside their job descriptions turn that experimentation into recurring routines. Following the first report’s finding of ‘task crossover,’ the new analysis looks at more than 1.5 million work-related ChatGPT messages from April through July 2026 and tracks roughly 6,200 workers over four months. The researchers find that cross-occupation AI use becomes a growing share of observed activity, that workers prompt AI differently for tasks outside their roles, and that some cross-occupation tasks stick far more than others—pointing to a path where jobs broaden in practice before titles change.

Workers prompt AI differently when the task is outside their usual role: prompts are shorter on average, and workers are less likely to ask for explanations, how-to guidance, a specific response format, or advice. They are more likely to provide examples or background and to ask AI to check or verify something. The authors interpret this as borrowing expertise: workers bring a problem and relevant context rather than asking to be taught a new field.

On recurrence, among the 6,200 consistently observed workers, previously used cross-occupation tasks rose from 13.1% of occupation-specific AI activity in April to 25.9% in July. A matched follow-up analysis found workers returned to a cross-occupation task used the previous month 23.6% of the time, versus 8.4% among comparable workers without prior use. Similar gaps appear for within-occupation and general tasks.

Recurrence varies considerably by task type: 54% for discussing goods or services with customers, 44% for advertising or promotional writing, and 37% for creating marketing materials, but only about 15% for explaining financial information. The average next-month return rate across cross-occupation tasks was 18.5%. The authors note the differences may reflect where AI fits naturally into workflows, or differences in workplace norms, caution, and the perceived consequences of getting something wrong.

The first two Work at the Frontier reports suggest a path for AI-driven job transformation: a worker experiments with an activity outside their traditional role, finds AI useful, and begins returning to it as part of their work. If those activities become regular responsibilities, the mix of activities within a job could broaden even while its title stays the same. The authors argue work design—how tasks are divided and carried out—deserves a place alongside access to AI tools in organizational AI strategy, and point to their AI Jobs Transition Framework for understanding how AI capabilities, human roles, and demand could shape employment.