43.5% of ChatGPT work tasks fall outside a user's job title, OpenAI finds

OpenAI’s economic research team dropped a new report that quantifies something many of us have sensed: AI is blowing up job descriptions. They analyzed over 800,000 messages from U.S. ChatGPT users and found that 16.8% of work-related chats and a striking 43.5% of occupation-specific messages involve tasks historically associated with a different job. The researchers call this “task crossover.”

This isn’t about AI replacing a worker. It’s about a small-business owner suddenly drafting a contract, or a marketer troubleshooting a busted website without pinging the developer. The study strips out generic stuff like scheduling to focus on specialized tasks. When they do, the crossover rates pop. A full 77% of occupation-specific messages from customer experience workers fall outside their traditional role. That figure is 75% for designers and 69% for HR. These people are “borrowing” work from other departments, reshaping the division of labor before any HR manager rewrites a job description.

The report, the first in OpenAI’s new “Work at the Frontier” series, maps where these tasks travel. Marketing and engineering tasks proved the most viral, frequently showing up in chats from people in other fields. Financial calculations and tech troubleshooting were the most common “outside” tasks across nearly every other occupation. Designers are heavy importers of outside work, with 35.2% of their messages borrowing from other jobs, while design tasks themselves rarely leak out. Engineering shows the opposite pattern, acting as a major supplier of tasks that others pick up.

Company size matters here. In a large org with specialized teams, this might cause friction. In a small shop where the person closest to the problem just needs to fix it, it’s pure leverage. The data suggests we’re watching a real-time reorganization of work, not a simple story of automation. As the researchers put it, AI changes “not just how work gets done, but who does what.”