Artificial intelligence news.
Straight from the original announcements.

OpenAI says AI is reshaping freelance work and project scope

November 10, 2026, 08:00Source: OpenAI

AI-related freelance projects are bringing together a broader mix of tasks and attracting larger fixed-price budgets, according to new research from OpenAI and Upwork. The findings suggest AI is changing both what clients hire freelancers to do and how they define projects.

Key points
  • 38% of skilled U.S. knowledge workers freelance, according to Upwork’s 2026 Future Workforce Index.
  • OpenAI analyzed 63 matched work subcategories across ChatGPT use and Upwork postings.
  • AI-related postings became 45% more diverse between early 2023 and late 2025.
  • AI-related projects had larger fixed-price budgets and involved more hours.
  • Clients requested AI alongside skills including video production and motion graphics.

Freelance postings offer an early view of AI’s impact

OpenAI studied Upwork job postings alongside activities people undertake directly with ChatGPT to examine how demand for freelance work is changing. The company says freelance work can reveal shifts sooner than economy-wide employment statistics because contracts are often short and clients can adjust requirements with each posting.

The research comes as freelance knowledge work represents a substantial part of the U.S. labor market. Upwork’s 2026 Future Workforce Index found that 38% of skilled U.S. knowledge workers freelance. Upwork’s previous annual study estimated that freelance knowledge work generated $1.5 trillion in U.S. earnings in 2024. Those estimates include work beyond Upwork’s own marketplace.

Demand is shifting across work categories

The types of work people undertake with ChatGPT do not match the services clients request from freelancers. Writing represented a much larger share of ChatGPT use, while video and animation represented larger shares of requests in Upwork postings.

Of the 63 work categories analyzed, 43 followed a pattern in which activities that were relatively common in ChatGPT use lost share among Upwork postings, while less common activities gained share. The other 20 categories showed the reverse pattern.

OpenAI cautions that ChatGPT use alone does not explain changes in freelance demand. The comparison reflects how people currently use ChatGPT, rather than the full range of tasks AI can support, a gap the company calls the capability overhang.

AI projects combine more capabilities and involve more work

By late 2025, clients were requesting AI applications across a wider range of freelance work, including chatbot development, video production, marketing automation and application integration. AI capabilities were often requested alongside other services within the same project.

One example cited by OpenAI involved a single posting seeking AI-generated video, editing, and motion graphics. In that case, a freelancer would combine those capabilities to deliver a finished video.

Between early 2023 and late 2025, the mix of tasks in AI-related postings became 45% more diverse, rising from about 15 equally common tasks to about 21. Posted hourly rates were similar for AI-related and other work, but AI-related projects had larger fixed-price budgets and involved more hours.

OpenAI says the findings do not point to a simple substitution story. Instead, they suggest that AI may be changing the activities clients seek from freelancers and the way they assemble projects, with implications that may appear in freelance postings before they become visible in broader employment data.

Related news

MIT finds enterprise AI adoption remains shallow despite rising agent use
ResearchNovember 10, 2026, 22:22

MIT finds enterprise AI adoption remains shallow despite rising agent use

Enterprise AI adoption remains far less widespread than headlines suggest, with only 11% of S&P 500 firms deeply integrating AI by the end of 2025, according to research…

GPT-6 Astra led coding agents, but finished only 26.4% of tasks
ResearchNovember 10, 2026, 17:08

GPT-6 Astra led coding agents, but finished only 26.4% of tasks

OpenAI’s GPT-6 Astra passed an average of 69.9% of hidden acceptance criteria in a benchmark of coding agents, yet fully finished only 26.4% of tasks. The Co-Design…

Fivetran study finds coding model price is a poor value guide
ResearchNovember 9, 2026, 17:02

Fivetran study finds coding model price is a poor value guide

A blinded Fivetran study of 39 engineers found that paying substantially more for an OpenAI coding model produced only a modest preference advantage in everyday software…