On October 22, 2026, Uber and OpenAI announced in an official joint statement that they would expand the ChatGPT Work pilot from 1,000 employees to company-wide deployment, making it available to over 30,000 employees across Uber globally. ChatGPT Work is located within ChatGPT Enterprise, and OpenAI defines it as an agent platform capable of handling complex, multi-step knowledge work. Official materials confirm that the coverage expansion began on that day, but do not indicate that all employees are actively using the service. No production, cost, or quality results were provided following the launch.
Expansion from Pilot to Global Employee Base
OpenAI referred to this expansion as Uber's first company-wide deployment of an agent platform designed for knowledge work. The official announcement specifically mentioned that operations, finance, marketing, and product functions would receive support, but did not disclose actual usage numbers by department. Therefore, the 'over 30,000 employees' should be understood as the global employee base eligible to access the service, rather than daily active users or those who have completed training. The initial pilot of 1,000 people was merely the officially disclosed scope number and cannot be used to infer pilot effectiveness.
According to official documentation, ChatGPT Work can connect to Google Drive, Gmail, Google Calendar, and Slack, assisting with research materials, preparing meeting briefings, creating presentations, analyzing data, and completing cross-application processes. OpenAI states that the agent can handle multi-step tasks and deliver completed products, but did not disclose what manual review processes these completed products undergo within Uber. These four are the connectors explicitly listed in the announcement and not a complete list of integrations. The materials also do not specify that the agent can read all of Uber's systems or all employee documents, with specific permissions still depending on approved integrations and access settings.
Connectors, Skill Library, and Training Boundaries
Uber plans to support its rollout with an internal AI skill library, approved integrations, and specialized training. This arrangement indicates that the company is not simply opening up a chat interface to employees, but rather incorporating tool selection, connection scope, and training into a unified deployment process.
The announcement does not list the contents of the skill library, duration of training, categories of data that can be processed, or procedures for handling violations. It also does not specify whether employees can independently add connectors. Since the deployment spans multiple functional departments, the same tools do not necessarily have identical access scopes; the announcement only confirms the existence of approved integrations without disclosing specific approval criteria, nor does it indicate whether training is a prerequisite for going live. Whether this can establish consistent working methods will depend on subsequent stages including employee adoption, permission configuration, task adaptation, and output review.
Default Model and Development Tool Division
OpenAI stated that GPT-6.1 Sol will become the default capability in ChatGPT Work deployed at Uber, and claimed that the model achieves a balance between performance and cost efficiency. The announcement also mentioned that Uber employees can utilize OpenAI's voice and image generation capabilities. The aforementioned product positioning and performance assessment come from OpenAI, while existing materials do not provide independent benchmarks, reasoning costs, or accuracy data specific to Uber's scenarios.
Software development and engineering teams will continue to use Codex to accelerate software development and code workflows. The official statement only confirmed that the tools will continue to exist, without providing metrics such as adoption rates, code acceptance rates, defect changes, or delivery cycle improvements. Neither did it state that Codex or ChatGPT Work would replace engineering positions. Positioning development tools and knowledge work agents side by side indicates that Uber has chosen a path of configuring different AI capabilities according to tasks.
Company Objectives vs. Media Narrative
Both companies described this expansion as a way to enable employees to handle more complex work, connect daily tools, and accomplish more tasks. Uber's Chief Technology Officer Praveen Neppalli Naga summarized the objective as providing employees with tools for innovation and problem-solving, while OpenAI President and Co-Founder Greg Brockman emphasized expanding the availability of cutting-edge models and agents. These are deployment objectives and corporate positions, not achieved business outcomes that can be written as implemented results.
TechCrunch's report published on October 22 at 16:00 UTC summarized the event as Uber making ChatGPT Work available to all 30,000 employees; however, OpenAI's official announcement used the phrasing "over 30,000 employees." There is a scope difference between these two formulations, and the count should follow the official wording, without extending the media headline's "all" to conclude active usage. Both sources confirmed company-level expansion, but neither disclosed actual adoption rates, and together they cannot prove usage effectiveness.
Organizational Governance and Verifiable Outcomes
The scale of deployment expands enterprise AI issues from individual assistants to cross-application workflows, unified models, connector approvals, and employee training. The broader the coverage, the greater the need for clear definitions of who can grant agents access to which data, who reviews the outputs, and how anomalous operations are tracked. While internal skill libraries and approved integrations provide governance entry points, they do not automatically equate to effective governance, especially when a single task spans email, calendar, documents, and chat records. The announcement outlines the organizational framework Uber intends to adopt, but does not disclose permission audit details and is insufficient to prove that privacy, security, quality, or return on investment have been independently verified.
The follow-up materials that can validate this deployment include actual adoption rates, common tasks, output quality, permission boundaries, and quantifiable business metrics. OpenAI and Uber have stated that they will share adoption status, use cases, and measurement results during the rollout process, but have not provided a timeline. The current accurate conclusion is that company-wide openness has been initiated, with employee coverage expanding from a 1,000-person pilot to over 30,000 people; any judgments regarding layoffs, cost savings, or productivity improvements still lack public evidence.