Does ChatGPT save voice recordings? What Voice, Live, and Dictation keep

The Morse teamPublished October 31, 20267 min read

Yes, for a while. As of October 2026, ChatGPT saves voice recordings from Live and Advanced Voice for 30 days alongside the chat's transcript, keeps Dictation audio for as long as the chat stays in your history, and deletes Standard Voice audio as soon as it's transcribed. OpenAI doesn't train on the audio unless you turn on "Include your audio recordings," but it can train on the transcripts whenever "Improve the model for everyone" is on.

We read OpenAI's ChatGPT Voice page (updated around October 21, 2026), its Dictation, Record, data controls, Temporary Chat, and file pages, its release notes, and its US privacy policy on October 23, 2026. OpenAI deserves credit for publishing a separate audio rule for each voice feature.

ChatGPT voice recordings at a glance

What is ChatGPT Voice now, and what happened to Advanced Voice Mode?

Advanced Voice Mode is now one of three options under Settings, then Voice. OpenAI's Voice page lists Live, powered by GPT-Live-1.1 or GPT-Live-1.1 Quip depending on your plan; Advanced, "the previous real-time Voice experience"; and Standard, which "transcribes your speech before generating a response."

OpenAI introduced GPT-Live-1 on July 8, 2026, and GPT-Live-1.1 on October 15, 2026 (release notes). The option you pick now decides what happens to your audio.

How long does ChatGPT keep your voice audio?

It depends on the feature, from deletion right after transcription to as long as the chat exists.

Some guides still say ChatGPT's voice audio is "stored along with chat history." OpenAI's Voice page now says clips from Live and Advanced are retained for 30 days. Dictation is the feature that keeps your audio with the chat.

Does OpenAI train on your ChatGPT voice recordings?

Not the audio, unless you opt in. OpenAI's Voice page says it won't train on audio or video clips unless you share them or enable the "Include your audio recordings" or "Include your video recordings" toggles. OpenAI's data controls page puts it this way: "Voice mode has a separate control for sharing data for model training."

Transcripts can still train models. The same Voice page says that if "Improve the model for everyone" is on, OpenAI "may use transcripts and other files from your Voice conversations to train our models." That setting is on by default for Free, Plus, and Pro, according to the data controls page.

If you do share clips, OpenAI says "our teams may review shared clips," for example where ChatGPT misheard you. Turning sharing off later stops new clips only: "Clips that were previously disassociated from your account may continue to be used." The Dictation page adds that one switch covers both, and turning it off "will stop sharing all audio across dictation and voice chat." For the training setting itself, see does ChatGPT train on your data.

What happens to voice clips when you delete a ChatGPT chat?

They go with it, within 30 days. OpenAI's Voice page says deleting a chat also deletes its audio and video clips "within 30 days," unless OpenAI needs them "for security, safety, or legal reasons" or you shared them for training and they were already disassociated from your account. Dictation audio follows the same rule, per the Dictation page.

Archiving isn't deleting. OpenAI says archiving "only removes the chat from your sidebar; it does not delete the chat or its associated audio or video clips." For when a court can make OpenAI keep chats longer, see the ChatGPT court order on data retention.

Does ChatGPT keep video and screen sharing from Voice?

Yes, for 30 days. Live "does not support video or screen sharing," so OpenAI's Voice page sends you to Advanced in the iOS and Android apps, where video clips are stored with the transcript for 30 days, like audio. A message preview or a bank balance on screen while you share is part of what you send.

Can you use ChatGPT Voice without it being saved?

Only partly. Standard deletes your audio after transcription, so pick it in Settings, then Voice, if stored audio is what worries you. The transcript still lands in your chat.

A Temporary Chat stays out of your history, isn't used for training, and "may be retained for up to 30 days for safety purposes," according to OpenAI's Temporary Chat page. As of October 23, 2026, neither that page nor the Voice page says how audio clips from a temporary chat are handled, so we can't tell you.

What does OpenAI's privacy policy say about audio?

It treats your voice as content, the same as a typed prompt. OpenAI's US privacy policy, updated September 10, 2026, lists "audio and video" among the "Content" it collects and gives no audio-specific retention period, only the general rule that deleted data goes "within 30 days" unless OpenAI must keep it longer. Its table of data categories for US state privacy laws doesn't list biometric information. The policy for the rest of the world, updated February 6, 2026, lists audio and video the same way.

How Morse keeps what you'd say out loud off your account

Whatever happens to the audio, every voice feature above ends with your words as text in a chat stored on OpenAI's servers, under an account with your email and payment details. A transcript of anything you'd say aloud is a chat message, and on Morse, chat history stays in your browser.

• The transcript problem. Your chat history is stored in your browser, encrypted, and locked when you sign out. Morse's servers don't keep your messages or log what you write, so there's no account history for anyone to review, and Morse usually has very little to give if a court asks.
• The two training switches. Morse doesn't train AI on your content, sell your data, or show ads. There's no "Improve the model for everyone" and no audio toggle to check.
• The 30-day wait. Morse's servers keep no copy, so deleting a chat or clearing your browser removes it from Morse, with nothing queued for deletion on our side. Morse can't recover it either, so keep what you need. Temporary chats are kept in memory only.

Then pick the mode for what you'd have said:

• Anonymous, to ask GPT what you'd have asked ChatGPT Voice, without the provider knowing it's you. Morse's proxy removes your account, email, and IP address and sends the message under Morse's own account. The provider sees the words and may keep them for a limited time to check for abuse, under its own policy, so leave out names and details that point to you.
• Private, the default, for questions no host should keep. Open-weight models like DeepSeek, Kimi, and GLM, plus xAI's Grok, with your identity stripped as in Anonymous. Morse has zero data retention agreements with its inference, search, and other providers, so the hosts agreed to keep nothing.
• E2EE, for anything no one else should read. Your browser checks the secure enclave's proof that it's genuine, then encrypts the message so only that enclave can read it, and Morse's servers relay ciphertext they can't read. It rests on the enclave hardware and attestation working as designed. Here's how end-to-end encrypted AI chat works.

Whichever mode you pick, our privacy policy lists everything the account holds: sign-in details, plan, balance, and a usage ledger with no chat content.

So what should you do about ChatGPT voice recordings?

• Stored audio: use Standard Voice, which deletes audio after transcription, and delete chats where you used Dictation, since that audio stays as long as the chat does.
• Training: leave "Include your audio recordings" and "Include your video recordings" off, and turn off "Improve the model for everyone" so transcripts stay out too. On Morse, there's no switch to find.
• Your history: delete, don't archive. On Morse, history sits only in your browser.
• Being identified: see how to use ChatGPT anonymously, or use Morse's Anonymous mode.

Try a chat from the Morse home page without an account, or start free with no card needed. End-to-end encrypted chats come with Pro, at $19 a month.