On October 6, 2026, OpenAI announced audio-file uploads in ChatGPT. Existing meeting, interview, or lecture recordings can be used to create transcripts, summaries, and answers about their contents.
Specify the output
Attach a supported audio file and explain the task. For example, ask for a topic timeline followed by decisions, action items, and unresolved questions. This is a suggested workflow, not a guarantee of perfect speaker identification or recognition of specialist terminology.
Different from a live call
Uploading an existing recording is distinct from having a real-time conversation in Voice or using other meeting-recording features. Access is available with paid subscriptions and workspaces, subject to model, region, app version, and administrative settings.
Replay consequential details
OpenAI warns that transcripts may contain errors and language performance can vary. Replay names, amounts, deadlines, and responsibility assignments before sending the summary externally. Also confirm that you are entitled to use and upload a recording containing other people.
