[macOS Desktop] Generic “chat” labels conflate Chat, Work, and Codex tasks

What version of the Codex App are you using (From “About Codex” dialog)?

26.825.41651 (build 7345)

What subscription do you have?

ChatGPT Pro

What platform is your computer?

macOS 26.6.2; Darwin 25.6.0 arm64 arm

What issue are you seeing?

The desktop app exposes Chat, ChatGPT Work, and Codex as different ways of working, but several controls and summaries collapse all of their objects into the generic term chat.

Observed examples on this build:

• The sidebar action remains New chat while Work is selected, even though the resulting object is a Work task.
• The Codex view also says New chat, although it creates a Codex task with a different execution context.
• Custom sections say Drop chats or projects here even when they contain Work or Codex tasks.
• Activity summaries can describe usage across a number of chats even when the activity includes Work and Codex tasks.
• Sidebar rows do not persistently identify whether an item is a Chat conversation, Work task, or Codex task.

This makes pinning, organizing custom sections, and reopening older work unnecessarily ambiguous. The user has to infer the object type from its location or manually encode it in the title.

This issue is specifically about incorrect control-level terminology and missing item-type identification. It is narrower than requests to redesign the entire sidebar or separate the product modes.

Official OpenAI documentation currently distinguishes Chat, ChatGPT Work, and Codex as separate choices:

https://learn.chatgpt.com/docs/use-chatgpt

What steps can reproduce the bug?

1. Open the ChatGPT desktop app on macOS.
2. Select ChatGPT, then select Work in the Chat/Work switcher.
3. Observe that the sidebar creation control still says New chat.
4. Switch the top-level product selector to Codex.
5. Observe that the creation control again says New chat, although it creates a Codex task.
6. Create or inspect a custom sidebar section and observe the instruction Drop chats or projects here for mixed Chat, Work, and Codex items.
7. Compare existing sidebar rows and observe that their originating item type is not persistently labeled.

What is the expected behavior?

Use terminology that preserves the actual object type:

• New chat for an ordinary Chat conversation.
• New Work task for ChatGPT Work.
• New Codex task for Codex.
• Neutral collection wording such as Drop conversations, tasks, or projects here where different object types are accepted.
• Persistent Chat, Work, or Codex badges/icons on sidebar rows and headers.
• Activity summaries that either preserve the item types or use a genuinely neutral term such as items.

The terminology should make it possible to identify an item's type without opening it or manually modifying its title.

Additional information

In-product feedback was submitted before this issue was filed.

Feedback ID: 01a05059-5756-7ef3-be56-9b492d17d106

Related issues:

• #33079 requests persistent Work/Codex identification and filtering.
• #33475 concerns ambiguity in the active execution mode.
• #33716 proposes separate organization models for Work and Codex.
• #35367 requests broader separation of ChatGPT, Work, and Codex projects and conversations.

This report is distinct because it focuses on the incorrect generic chat microcopy used by creation controls, section instructions, activity summaries, and sidebar item labels.

Diagnostics: NONE
Logs or traces: NONE
Screenshots or attachments: NONE

No files or tests are named. Start by locating the desktop UI strings for “New chat,” “Drop chats or projects here,” activity summaries, and sidebar rows, then trace how Chat, Work, and Codex modes identify their objects. Done means controls and summaries use accurate or neutral terminology and sidebar items persistently identify their type.