What changed
OpenAI’s October 6 release notes add audio uploads to ChatGPT. A user can attach a supported recording and ask for a transcript, summary, answers about its contents, structured notes or a follow-up email draft. The practical difference is that the audio itself becomes a first-class input rather than requiring a separate transcription step first.
OpenAI lists meetings, interviews and lectures as examples. A useful workflow is to ask for decisions, open questions and named owners, then verify those details against the source audio before sending anything. The model can also help search the recording conversationally: ask what a participant said about a deadline or which options were discussed.
What the result actually shows
There are limits. Audio uploads are available with paid subscriptions and workspaces, and availability can vary by region, workspace settings, client version and model. OpenAI also warns that transcripts may contain errors and performance varies across languages.
So this is a convenient summarization path, not a guaranteed verbatim record. For high-stakes minutes, preserve the original recording, check names and numbers, and treat any generated follow-up as a draft.
Source published 2026-10-06. Coverage is based on the maker’s announcement and demonstration.