OpenAI published new Work at the Frontier research on July 27 arguing that ChatGPT is changing not only how people do work, but which tasks they take on.
The company says its analysis of more than 800,000 U.S. ChatGPT messages found that 16.8% of work-related messages and 43.5% of occupation-specific messages are about tasks associated with another occupation. OpenAI calls this pattern task crossover: work historically associated with one role appearing in the AI use of someone in another role.
That is a more specific claim than “AI boosts productivity.” It suggests people are using AI to cross role boundaries before companies formally rewrite job descriptions, reporting lines, or team structures.
Some roles borrow more work than others
OpenAI says the outside-occupation pattern is especially visible once generic tasks such as writing, summarizing, and scheduling are excluded.
Among occupation-specific messages, OpenAI reports outside-occupation task shares of 77% for customer experience workers, 75% for designers, 69% for human resources workers, 56% for legal workers, and 53% for marketers. The company also says marketing and engineering tasks travel farthest across occupations in its heatmap.
The examples are practical. A small-business owner may draft copy, review a contract, or perform basic financial analysis. A salesperson may explore a customer dataset. A marketer may troubleshoot a website without waiting for a developer.
Usage data is not the same as labor-market proof
The finding is useful because it looks at what people ask an AI system to help with, not only what economists assume could be automated from a task list.
It also has limits. ChatGPT usage does not prove that workers completed the task well, that employers changed roles, or that wages and headcount moved because of the tool. The data shows demand for cross-functional help inside ChatGPT conversations. It does not by itself show durable job redesign.
Still, the signal is worth tracking. If people routinely use AI to take on work that previously required another function, managers may start reorganizing around problems and outcomes rather than strict role boundaries. That change could be most visible in smaller organizations, where there are fewer specialist teams to hand work to.
