OpenAI published a report on September 16, 2026, part of its Work at the Frontier research series, examining how workers use AI for tasks outside their typical occupations. The company analyzed more than 1.5 million work-related ChatGPT messages sent between April and July 2026. According to OpenAI, workers prompt AI differently depending on whether a task fits their usual role: for tasks outside their occupation, they write shorter prompts on average, are less likely to ask for explanations or how-to guidance, and are more likely to supply examples or background and to ask AI to check or verify something. OpenAI interprets this as workers 'borrowing expertise' by bringing a problem and context rather than learning a new field. Among roughly 6,200 workers observed consistently, previously used cross-occupation tasks rose from 13.1% of occupation-specific AI activity in April to 25.9% in July. In matched one-month follow-ups, workers returned to a prior cross-occupation task 23.6% of the time, versus 8.4% for comparable workers who had not used it. Return rates varied by task type: 'discussing goods or services with customers' recurred at 54%, while 'explaining financial information' recurred about 15%, with an average of 18.5%. OpenAI says jobs may broaden even while titles stay the same.
- Study analyzed 1.5 million+ work-related ChatGPT messages from April–July 2026
- Cross-occupation tasks rose from 13.1% to 25.9% of occupation-specific AI activity among ~6,200 workers
- Prior cross-occupation tasks recurred 23.6% of the time vs. 8.4% for non-users; average return rate 18.5%
What it means for you
OpenAI's own research says people are using AI to handle tasks that normally sit outside their job — and doing it repeatedly, not just once. In plain terms: an AI tool lets one person cover ground that used to need a specialist, and those borrowed tasks tend to stick. It's an interesting trend report, but it doesn't hand you a new tool or a decision to make.
Try this
Pick one recurring task in your business that currently gets outsourced or bottlenecked on one person — a customer email, a rough financial summary — and try handing it to ChatGPT with a real example and context, then verify the output yourself before using it.
Who should care
Small-business owners and team leads thinking about how AI might broaden what individual staff can do without hiring specialists.
Skip this if
You're not actively rethinking how work is divided on your team, or you just want to know about new product capabilities — this is a labor-trends study from a vendor, not a tool release.
Sources: OpenAI — read the original