Editorial illustration for OpenAI Report Finds Only 30% of ChatGPT Use Is for Work
Only 30% of ChatGPT Use Is Work-Related, OpenAI Finds
OpenAI released new figures this month showing that only 30% of ChatGPT conversations involve work-related tasks, a number that complicates the narrative AI companies have been selling about productivity gains and enterprise adoption. The rest, apparently, is personal: advice, companionship, curiosity, whatever else people type into a chat box at 11 p.m.
The catch is that OpenAI produced this data itself, and so did Anthropic when it published similar usage reports on Claude. Both companies choose what to release and how to frame it, which leaves researchers, regulators, and journalists working from numbers they can't independently check. That gap is what a Stanford-led project called the AI Observatory is trying to close. Led in part by Anka Reuel, a PhD candidate at the Stanford Trustworthy AI Research Lab, the project pulls together real conversations with models like Claude and Gemini, collected with user consent across seven separate datasets, to build a public record that doesn't depend on what a company decides to disclose.
The stakes go beyond curiosity. Governments and companies are making policy calls about AI's risks and benefits based on data nobody outside the labs can verify.
The AI Observatory found that AI use differs significantly across models, and has changed over time. Its research shows many more sensitive behaviors than are captured in reports from major AI companies, which they say focus more on work than on personal use.
Why this matters
For developers and founders building on top of ChatGPT or Claude, the 30% figure is worth sitting with. Most of the roadmap conversation in this industry still assumes AI is primarily a productivity tool, something bolted onto workflows to make knowledge work faster. OpenAI's own numbers say otherwise: seven in ten consumer chats have nothing to do with work at all.
That's a different product than the one investor decks describe. Anka Reuel's point about independent corroboration matters here too. Anthropic and OpenAI decide which slices of usage data to publish, and there's no outside audit checking those numbers against what's actually happening on their servers.
When Anthropic writes about companionship use or CSAM generation in a blog post, that's the company grading its own homework. Researchers building products, safety tools, or policy recommendations on top of these self-reported stats are working with a picture the labs chose to show them. Until someone outside Anthropic and OpenAI gets real access to usage data, "how people actually use AI" remains a claim, not a fact.
Related Reading
Browse all OpenAI news →
LLMs & Generative AI ChatGPT's 'Nerdy' tweak rewards goblin metaphors in answers, study finds
LLMs & Generative AI Google tests visual 'magazine-style' UI for Gemini 3 Pro users
LLMs & Generative AI AI Engineers Face Rising Costs, Need New Strategies for Efficiency
Related Topic OpenAI's GPT Transcribe Cuts Error Rate to 3.31%, Improving on GPT-4o
Open Source
Related Topic OpenAI Calls Apple's Lawsuit 'Rotten to Its Core' in Rebuttal
Policy & Regulation
From Archives AI Compressors Drop 83% of User Instructions, Study Finds
From Archives Researchers Build 128-GPU Cluster From Scrap, Run LLaMA-70B for a Year
From Archives OpenAI Dissolves AI Risk Team, Splits Duties to Existing Divisions
Open Source
From Archives OpenAI Shuts Down Team Tasked With Evaluating AI Catastrophic Risks
Open Source
Common Questions Answered
What percentage of ChatGPT conversations are work-related according to OpenAI's report?
According to OpenAI's report released this month, only 30% of ChatGPT conversations involve work-related tasks. This means that 70% of consumer chats are used for personal purposes such as advice, companionship, and general curiosity rather than productivity-focused work.
Why is the 30% work-usage figure significant for developers building on ChatGPT or Claude?
The 30% figure challenges the dominant narrative in the AI industry that positions these tools primarily as productivity solutions for knowledge work. Since seven in ten consumer chats have nothing to do with work, developers and founders need to reconsider their product roadmaps and recognize that the actual use case differs substantially from what investor decks typically describe.
What does the AI Observatory say about the usage data reported by major AI companies?
The AI Observatory found that AI use differs significantly across models and has changed over time, revealing many more sensitive behaviors than are captured in reports from major AI companies like OpenAI and Anthropic. The Observatory suggests that these companies' reports focus more heavily on work-related use than on personal use, potentially skewing the perception of how people actually utilize AI tools.
How do OpenAI and Anthropic's usage reports compare when it comes to transparency?
Both OpenAI and Anthropic have published similar usage reports on their respective AI tools, but there is an important caveat: both companies produced this data themselves. This self-reported nature of the data raises questions about independent corroboration and the potential bias in how these companies present their usage statistics.
Further Reading
- How AI is expanding what people do at work - OpenAI - OpenAI
- How people are using ChatGPT - OpenAI - OpenAI
- This month in AI: ChatGPT usage patterns, governance ... - World Economic Forum
- OpenAI says more workers are using ChatGPT to do other people's jobs - The Decoder
- OpenAI's New Report Details How We Use ChatGPT at Work - Marketing AI Institute
