OpenAI Says a Quarter of U.S. Workers Now Use ChatGPT on the Job

More than a quarter of U.S. workers — and 45% of those with postgraduate degrees — now report using ChatGPT for work, according to a new OpenAI report on workplace AI adoption. The figure anchors "Inside GPT-5 for Work," OpenAI's data-driven account of how its chatbot has moved from a research curiosity to a fixture of daily work in under three years.

The scale is unusual for enterprise software. Since its November 2022 launch, ChatGPT has grown from a tool aimed at a small group of AI researchers and enthusiasts to more than 700 million weekly active users, making it one of the world's most visited websites. It reached 100 million weekly active users within months. OpenAI attributes its workplace foothold to a bottom-up adoption pattern: employees carried the product in from their personal lives, skipping the long decision cycles, training programs, and top-down rollouts that typically gate enterprise software. The company calls it "the fastest adopted enterprise technology in recent history."

The stakes for the report go beyond OpenAI's own product metrics. The document combines OpenAI's analysis of anonymized and aggregated ChatGPT and ChatGPT Enterprise usage data — processed with automated content classifiers, with no review of user or customer content — with independent third-party studies, including peer-reviewed research. It arrives as employers, regulators, and rivals are all trying to gauge how deeply generative AI has actually penetrated the workforce, and in which sectors the technology is delivering measurable returns versus hype.

Adoption is uneven across industries

IT and finance lead adoption, which OpenAI says tracks with the tool's strengths in coding, analysis, and information-heavy work. Manufacturing shows higher-than-expected uptake, which the company reads as a sign of broader digital transformation: factories applying AI to process automation, predictive maintenance, and supply chain optimization, with early industrial AI investments paving the way for engineers, analysts, and operations managers to use ChatGPT routinely.

Other sectors lag. Retail, construction, transportation, wholesale trade, and agriculture all show significantly lower adoption, which OpenAI attributes to their smaller share of knowledge workers. Healthcare is a special case: despite being one of the largest and most data-intensive industries, adoption has been slower, a pattern the report links to strict privacy and compliance rules and risk-averse organizational cultures. Still, OpenAI reports growth in targeted areas like clinical documentation and administrative workflows, and suggests healthcare could soon become a hotbed of AI adoption.

Four tasks dominate early usage

In users' first three months with ChatGPT Enterprise, four categories dominate: writing, research, programming, and analysis. Together they account for the majority of messages sent, according to aggregated department data collected during onboarding and classified automatically.

Technical teams are among the heaviest users, with analytics, engineering, and IT roles making up a large share of early usage. Programming is the top task for engineering roles, but OpenAI notes substantial demand for research and documentation help as well — suggesting ChatGPT is used nearly as much for planning as for coding. IT teams lean most heavily on research and troubleshooting, treating the product as an information resource before moving into automation.

Go-to-market functions — marketing, communications, sales, and customer experience — are also major adopters, relying on ChatGPT primarily for writing, research, creative ideation, and media generation. Design teams stand out for media generation, using it 2–4x more than other groups.

One notable finding: coding is spreading beyond engineering. Designers use ChatGPT for programming at a much higher rate than finance and sales, likely for front-end prototyping and snippet help. Project managers combine writing, media generation, coding, and data analysis. Product, operations, marketing, finance, and HR all use it for coding to some extent.

Independent evidence of skill gains

OpenAI points to external validation. A study by Boston University and BCG examined ChatGPT's impact on the technical competency of BCG consultants. Consultants armed with and trained on ChatGPT scored 49, 20, and 18 percentage points higher than a control group on three technical tasks, and performed close to the level of real BCG data scientists on two of the three tasks.

Two other studies cited in the report address productivity and job satisfaction directly. In a six-month randomized field experiment across thousands of knowledge workers, access to AI cut weekly email time by 31%. A separate study of software developers found that AI coding tools let them spend more time coding and on exploratory work, and less time on project management. Together, OpenAI argues, the findings suggest tools like ChatGPT reduce busywork and free time for more strategic, higher-value work.

Advanced features remain underused

Most departments rely on ChatGPT's core tools — search, data analysis, file uploads, retrieval, and canvas. More advanced capabilities, including reasoning models, deep research, projects, and custom instructions, see higher usage mainly among power users such as R&D teams. OpenAI identifies this gap as a missed opportunity, with barriers that may include discoverability, awareness of use cases, or the setup required.

GPT-5 is OpenAI's answer to part of that problem. Its real-time router automatically decides which advanced features and tools to use based on conversation type, complexity, tool needs, and explicit intent.

Usage patterns also diverge by team. IT teams favor retrieval and search, treating ChatGPT as a knowledge companion for configuration and policy questions. Engineering teams show stronger use of GPTs, programming tools, and data analysis. That divergence, the report notes, shows adoption depends not only on technical fluency but also on the type of work and context within each department.

Usage intensity is climbing alongside breadth. Certain power-user segments of ChatGPT Pro subscribers send upwards of 200 messages per day, and usage has evolved from simple Q&A to coding, data analysis, and agentic workflows.

From tool to platform

OpenAI frames the next phase as a shift from personal productivity to workflow infrastructure. New capabilities — autonomous agents, advanced coding support, decision-assist tools — are expanding ChatGPT beyond individual tasks. "Increasingly, ChatGPT functions as an operating system for daily work: a shared layer where decisions are made, problems are solved, and output scales," the report states. Executives use it to shape strategy, engineers to design and debug systems, and customer support agents to evaluate complex solutions.

The company also argues the technology reshapes job design itself. Because ChatGPT reaches across functions, individuals can take on tasks once spread across multiple departments. OpenAI's example: a product manager might use it to analyze customer feedback, test and refine a new feature, and draft the legal and marketing content needed to bring it to market. Features like memory are making the product more context-aware, and the ability to bring structured and unstructured data directly into ChatGPT is broadening its role as the central interface for enterprise knowledge — a shift OpenAI says GPT-5 is accelerating.

The report closes with a historical analogy: electricity reshaped factory work, the internet redefined commerce and communication, and AI is setting the stage for the next leap. OpenAI's bet is that as adoption moves from low-friction daily tasks toward deeper reasoning and collaborative workflows — driven by enablement programs and product improvements that lower barriers — the enterprises that adapt quickly will capture faster decision cycles, productivity breakthroughs, and gains across every function. The usage data suggests most companies are still early in that curve, and that the largest changes in how work gets organized are still ahead.