New joint research by OpenAI and Upwork shows that AI is reshaping demand for freelance services, the composition of projects, and budget patterns. Some traditional service areas have lost marketplace share while AI-related postings bring together varied skills and tend to have larger fixed-price budgets and more total hours of work.
Why this matters
Upwork’s 2026 Future Workforce Index found that 38% of skilled U.S. knowledge workers freelance. Upwork’s prior annual estimate put freelance knowledge work earnings in the U.S. at about $1.5 trillion in 2024. These figures extend beyond Upwork’s own marketplace and underline how significant freelance work is to the broader economy.
Because freelance contracts are often short and client requirements can change quickly, freelance postings and ChatGPT usage can reveal changes in work more rapidly than aggregate employment statistics.
Data sources and timeframes
- ChatGPT usage: January 2025–June 2026.
- Upwork postings: analyses reference H1 2026 and comparisons across periods including H1 2023 to H2 2025.
- The study matched 63 Upwork work subcategories to analyze patterns across tasks.
Key findings
-
Many activities commonly performed in ChatGPT lost share on Upwork, while other, less common ChatGPT activities gained share on Upwork. Of the 63 matched subcategories, 43 followed this pattern and 20 showed the opposite.
-
By late 2025, clients requested AI to be used in many more and different kinds of work, including chatbot development, video production, marketing automation, and application integration.
-
AI capabilities were often requested alongside other tasks within single projects. AI‑related postings listed roughly as many tasks as other postings but more frequently combined tasks that had rarely appeared together before. For example, some postings asked for AI‑generated video, editing, and motion graphics in one package.
-
The mix of tasks in AI‑related postings became 45% more diverse between early 2023 and late 2025 — roughly equivalent to moving from about 15 equally common tasks to about 21.
-
Posted hourly rates were similar between AI‑related and other work, but AI‑related projects had larger fixed‑price budgets and involved more hours. This pattern suggests that higher budgets may partly reflect a broader scope of work rather than higher hourly pay.
Limits and interpretation
The comparison does not support a simple substitution narrative in which AI uniformly replaces human freelancers. While some activities common in ChatGPT decreased in Upwork share, the pattern was not uniform across categories. ChatGPT usage data also reflect only how people currently use the tool, not the full set of tasks AI could eventually support (a gap OpenAI terms the "capability overhang").
Freelance postings turn over quickly, and work inside firms can evolve without new job listings. Together with ChatGPT data, these postings provide an early look at which activities people perform directly with AI, which they contract out to freelancers, and how AI‑related projects differ from other freelance work.
Conclusion
The joint OpenAI and Upwork analysis indicates that AI is changing both what clients seek from freelancers and how projects are assembled: expanding task diversity, producing new task combinations, and increasing fixed‑price budgets and total hours for AI‑related work. The OpenAI Economic Research Team continues to explore these dynamics in its Work at the Frontier series.
