openai / openai/codex

[ChatGPT Desktop][macOS][Realtime Voice] Says requested text was shared, but no message appears

Open
#41,832 2 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

app bug model-behavior
Dominant language
Rust
Stars
125k
Forks
19.4k
PR merge metrics
PR metrics pending

Description

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

Not captured for this reproduction.

What subscription do you have?

ChatGPT Pro ($200/month)

What platform is your computer?

macOS 26.6.2 (Darwin 25.6.0 arm64 arm)

What issue are you seeing?

During a Realtime Voice conversation, I asked the assistant to place a substantial prompt or other text into the visible chat so I could copy it.

Voice stated that the requested text had been provided or shared in the chat, but no corresponding message containing that text appeared. When challenged, Voice repeated that it had already provided the information instead of recognizing that no visible message had been created.

This is not merely a formatting problem: the requested substantive content is absent while Voice reports the delivery as complete.

What steps can reproduce the bug?
  1. Start a Realtime Voice conversation in the macOS desktop app.
  2. Ask Voice to write a substantial prompt or other text into the visible chat so it can be copied.
  3. Wait for Voice to say that it provided or shared the requested text.
  4. Inspect the visible conversation and observe that the requested text is missing.
  5. Ask Voice where the text is and observe that it may repeat the completion claim even though no corresponding message exists.
What is the expected behavior?

When Voice says requested text was shared in the chat, the complete text should appear as a durable, visible message in the same conversation.

Voice should verify that the message was actually rendered before claiming completion. If it cannot create the visible message, it should clearly disclose that limitation and provide an available alternative instead of saying the text was already shared.

Additional information

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

Feedback ID: 01a05246-22fc-7fe1-88e9-fb1821b2d363

Related but not exact duplicates checked before filing:

  • #39472 concerns Voice claiming live access to another chat without tool evidence.
  • #37900 concerns Markdown links disappearing from an otherwise visible Voice transcript.
  • #39686 concerns both user and assistant transcripts not rendering on Windows.
  • #41254 concerns successfully delivered Voice turns remaining in a failed Retry queue.
  • #39172 is a feature request about choosing what Voice speaks versus displays.

This report is distinct because Voice specifically claims that requested substantive text was placed in the current chat, while no message containing that text exists.

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

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

No source files or tests are identified. Reproduce the issue in the macOS desktop app during a Realtime Voice conversation, then trace how the claimed shared text is created and rendered in the current conversation. Done means the complete requested text appears as a durable visible message, or Voice clearly reports that it could not create one.

Written by the indexing model from the issue text.

Assessment

Tech stack
macos
Domain
audio-video-rtc, desktop
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Active
Clarity
Mostly clear
Newbie friendliness
42/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.