OpenAI said on July 9, 2026 that ChatGPT Work runs in the cloud on web and mobile, while the new desktop app can use local files and desktop apps with user permission. The official release notes also say the new macOS and Windows desktop app combines Chat, Work, and Codex in one app, with Work rolling out first to paid plans except Free and Go. For practitioners, this is a permission and data-flow story: workflows may cross browser, cloud, local files, desktop apps, and connected plugins, so enterprise teams need clear policies for allowed actions, local-file access, and audit trails.
The important point for builders is the split between cloud agent work and desktop-assisted local work. That split affects what data leaves a device, what permissions users grant, what logs administrators can review, and how much trust teams should place in local app access versus cloud execution. What happened: OpenAI's July 9, 2026 ChatGPT release notes introduced ChatGPT Work and described it as an agent for longer tasks that can research, analyze information, work across connected apps and files, and create finished documents, spreadsheets, presentations, reports, and Sites. The same release notes say Work runs on web and mobile for paid plans except Free and Go, and that the new ChatGPT desktop app for macOS and Windows combines Chat, Work, and Codex. Technical context: OpenAI says Work on desktop can use local files and desktop apps with user permission, while web and mobile Work runs in the cloud. For enterprise teams, that creates different control surfaces: browser/cloud workflows need workspace-level app and plugin permissions, while desktop workflows also need local-file, local-app, and endpoint-management policies. The release notes also say existing app connections are unaffected as the App Directory becomes the Plugin Directory. For practitioners: Treat this as an access-control and auditability update before treating it as a pure productivity launch. Teams should decide which connected apps Work may touch, which local file classes are allowed, when approval is required before changes, and how to preserve review logs for generated documents, code edits, and cross-app actions. What to watch: Watch admin controls during the Enterprise and Edu preview, how public Sites publishing is governed, and whether OpenAI publishes more detailed desktop-local execution guidance. The key operational signal will be whether organizations can enforce policy consistently across web, mobile, and desktop Work sessions.
Key Points #
- 1OpenAI's official notes distinguish cloud Work on web/mobile from permissioned local access in the desktop app.
- 2Combining Chat, Work, and Codex changes enterprise controls for files, apps, browser actions, and generated deliverables.
- 3Practitioners should review connected-app policies, local-file permissions, approval gates, and audit logging before broad rollout.
Scoring Rationale #
This is a product-control and data-flow update with practical implications for organizations adopting ChatGPT Work and Codex. It is moderately important for enterprise operations, but it is not a standalone model release or broad market shock.
Sources #
Public references used for this report. Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.