Uber Expands ChatGPT Work Access to Over 30,000 Employees
Uber began a company-wide ChatGPT Work rollout on October 22, expanding from a 1,000-employee pilot to access for more than 30,000 staff. OpenAI describes approved integrations and training, with GPT-6.1 Sol as the default capability. The announcement promises later outcome reporting but contains no company-wide productivity result.
Uber expands ChatGPT Work beyond a pilot
OpenAI said on October 22 that Uber was starting a company-wide rollout of ChatGPT Work inside ChatGPT Enterprise. The stated audience is more than 30,000 employees worldwide, up from a pilot of 1,000. These are access and rollout figures supplied by the companies, not a count of daily users or completed tasks. The source calls this Uber's first company-wide deployment of an agentic platform. This article appears in the October 26 issue and retains the announcement's original October 22 date.
ChatGPT Work is described as an agent for multi-step work: employees can research information, prepare meeting briefs, create presentations, analyze data and complete workflows. The announcement names Google Drive, Gmail, Google Calendar and Slack as connected tools. A connector can make a workflow useful only if its authorization matches the employee's actual rights. The source does not describe the exact permission matrix, retention configuration or approval steps for every tool action, so readers should not infer that all 30,000 staff can use every integration in the same way.
OpenAI says GPT-6.1 Sol will be the default model capability in ChatGPT Work for Uber. The statement concerns the product configuration described for this rollout; it is not a measured speed or quality result. Uber says an internal AI skills library, approved integrations and employee training will support workers in operations, finance, marketing and product. Its developers and technical teams have separate ways to use OpenAI systems. A company-wide access decision does not by itself imply that every team has completed training or moved a critical process to the agent.
Adoption is different from outcome
The shift from 1,000 pilot participants to over 30,000 eligible employees makes this a noteworthy enterprise deployment. But the original announcement does not report hours saved, error rates, employee satisfaction, incident counts or business returns for the expanded population. The companies say they plan to share adoption, use cases and measurable outcomes as rollout progresses. That promise is a future reporting intention rather than current outcome evidence.
The most useful questions for an enterprise rollout are concrete. Which tasks can the agent finish without approval? Which files and conversations can its connectors read? How are summaries checked against their sources? Who owns a failed action, and can a worker see what the agent changed? A pilot may reveal common use cases, but scaling access by a factor above 30 changes support, auditing and training needs. That factor compares announced headcounts and is not an efficiency gain.
Uber and OpenAI had described earlier cooperation on customer-facing and driver-facing use cases. This announcement has a different subject: internal employee access to ChatGPT Work. Readers should not merge those programs into one performance claim. Evidence from a customer feature cannot automatically validate an internal finance or operations workflow, and a staff rollout says nothing about the quality of a consumer product.
What to watch after rollout
A credible later update would separate eligible seats, active users and repeated useful use. It would report task-level comparisons that include human review time, corrections and permission failures, not just the number of prompts sent. It would also describe which integrations were approved and how sensitive information is handled. These are editorial criteria; the announcement does not supply the figures yet.
For now the defensible conclusion is narrower: Uber has announced a large expansion of access to an agentic work product, supported by training and controlled integrations, and says it will publish more measurable outcomes later. The source is OpenAI's company announcement on October 22. The October 26 issue date does not turn a planned result into a measured one.
Sources
FAQ
How large is Uber's ChatGPT Work rollout?
OpenAI says Uber is expanding access from a pilot of 1,000 employees to more than 30,000 employees globally. The October 22 announcement calls it Uber's first company-wide agentic platform deployment. These numbers describe pilot size and eligible rollout audience, not daily active users, completed tasks or measured savings.
Which tools and model are described for Uber employees?
The announcement names Google Drive, Gmail, Google Calendar and Slack integrations and says GPT-6.1 Sol will be the default model capability in ChatGPT Work. Uber also plans approved integrations, an internal AI skills library and training. The source does not publish a permission matrix for every employee or task.
What evidence is still missing after the access expansion?
The companies say they intend to share adoption, use cases and measurable outcomes as rollout progresses. The initial statement does not report company-wide error rates, time savings, employee satisfaction or returns. Readers should distinguish eligible seats from active users and compare completed work including human review and corrections.