How to Use ChatGPT Temporary Chat: What It Protects and What It Does Not

ChatGPT Temporary Chat lets you have a conversation that is not added to your ordinary history unless you choose to save it. Before the first message, the current documented interface lets you choose Personalized or Unpersonalized; neither creates or updates memories while the conversation stays temporary. Temporary chats are not used for model improvement, but OpenAI may retain a copy for up to 30 days for safety.

What Temporary Chat means in ChatGPT

A temporary chat is a conversation with different history and data-use behavior from an ordinary saved chat. It is useful for a one-off explanation, a short brainstorming session, or an experiment you do not want filling your sidebar. It is still an interaction with an online service, not a private notebook that only exists on your device.

The details in this guide were checked against OpenAI’s Temporary Chat documentation on November 2, 2026. The page was updated by OpenAI within the past week. Current documentation includes personalization, saving, and file-retention details that some older tutorials do not describe. If your account shows a different rollout, use the options actually displayed and check the official instructions before proceeding.

Two questions are worth separating: “Will this conversation appear in my normal history?” and “Can this conversation use information already saved about me?” A temporary conversation can avoid ordinary history while still using existing personalization if you choose that mode. That is why the choice before your first message matters.

Temporary does not mean suitable for every sensitive task. Employer rules, workspace restrictions, third-party services, and information about other people still matter. You can learn the controls using a harmless fictional example before deciding whether the mode fits your real task.

How to start a ChatGPT Temporary Chat

Begin with a new conversation rather than assuming an existing chat has changed modes. Look for the Temporary control in the new-chat interface. Placement can differ between web and mobile versions, so read the mode indicator before entering information.

1. Open ChatGPT and check your account and workspace.
2. Start a new chat.
3. Select Temporary.
4. Choose Personalized or Unpersonalized, if that choice is available.
5. Check the visible temporary-chat indicator.
6. Enter a low-risk request and review the response.

The current official instructions say you cannot change the personalization choice after the conversation starts. If you chose the wrong mode, start a new temporary conversation with the intended setting. Do not rely on a later prompt such as “ignore everything you know about me” as a substitute for the account control.

On mobile, use the app’s new-chat and temporary-chat controls rather than copying a desktop shortcut from an old screenshot. If the option is unavailable in a work or school workspace, ask the administrator about restrictions. Creating a personal account to upload restricted workplace material would not solve the underlying permission issue.

Before leaving a useful temporary conversation, decide whether you want to keep anything. Unsaved temporary chats do not appear in ordinary history and cannot be reopened later through that history. Saving changes their status, as explained below.

Personalized versus Unpersonalized temporary chats

Personalized temporary chat can use existing memories, custom instructions, and plugins. Unpersonalized temporary chat does not use those sources. Neither mode creates or updates memories while the chat remains temporary.

Before the conversation begins, choose whether the temporary chat is personalized. A personalized temporary chat can use existing memories, custom instructions, and plugins, subject to your account and workspace settings. You can turn personalization off when you start the chat.

Source: OpenAI, Data controls in ChatGPT, checked November 2, 2026.

That choice is made per conversation, and it is easy to miss because it appears before your first message rather than in Settings. If you want a temporary chat that starts from a blank slate, look for the toggle at the point you open the mode.

Choose Personalized when a harmless one-off task benefits from preferences you already supplied, such as the reading style you normally use. Choose Unpersonalized when you want a response without those existing preferences. For example, an unpersonalized comparison of two fictional invitations can help you see how your current instructions work on their own.

Unpersonalized does not promise that every form of context is technically impossible. OpenAI describes limited safety-relevant context in rare, high-risk situations as separate from personalization. Workspace restrictions continue to apply regardless of the mode you choose.

Memory itself is not a transcript of everything you have ever said. OpenAI’s memory guide explains the available memory sources and management controls. For a full walkthrough of what Memory stores and how to delete it, see the ChatGPT Memory guide. If your goal is to remove existing remembered information, manage those sources directly rather than starting a temporary chat and assuming the old information has been erased.

Does ChatGPT save your data in Temporary Chat?

OpenAI’s documentation lists what is true of a chat while it remains temporary:

While they remain temporary, these chats:

• do not appear in your chat history
• do not create or update memories
• are not used to improve OpenAI models
• may be retained for up to 30 days for safety purposes

Source: OpenAI, Data controls in ChatGPT, checked November 2, 2026.

Those four lines answer most of the question, and the last one is the one people skip. Limited retention exists even when ordinary history and training use are switched off. Temporary is not the same as deleted.

An unsaved temporary chat does not become a normal searchable item in your account history. That makes it useful for sidebar clutter, but it also means you should not depend on finding it tomorrow. A conversation that matters for a project needs an intentional storage decision.

The phrase “does ChatGPT save your data” therefore needs a more precise answer than yes or no. Ask whether you mean normal chat history, safety retention, saved memory, a file stored elsewhere, or information sent to another service. Temporary Chat changes some of these behaviors without turning off every possible form of processing.

For a personal example, asking for five names for a fictional knitting club is a reasonable way to explore the mode. Pasting a bank statement, a relative’s medical history, or a confidential client file introduces data concerns that the word temporary does not resolve. Use the least information needed for the task.

Saving a temporary chat changes the rules

Current OpenAI documentation says you can save a temporary conversation as an ordinary chat:

If you save a temporary chat, it becomes a regular chat in your history and follows your account’s personalization and model-improvement settings.

Source: OpenAI, Data controls in ChatGPT, checked November 2, 2026.

This is a meaningful change of status. The protection described as “not used for model improvement while temporary” should not be carried over as an unconditional promise about a conversation you later save. Review the setting before treating Save as a harmless bookmark.

If you only need a short nonsensitive result, you can choose to copy the relevant text into an appropriate document instead. That creates a local or cloud copy under the rules of that storage system. It does not convert the original conversation into ordinary history, and it does not erase any safety retention already described by OpenAI.

A useful decision is to ask what you need to preserve: the answer, the full exchange, or nothing. Keeping a short grocery idea is different from preserving a business record. Avoid saving more private context than the result requires.

If your temporary chat includes uploaded files, saving the chat may also save eligible files to Library or Space, subject to availability and storage limits. Files saved separately in Library or Space follow their own retention rules and do not disappear when a temporary chat is left unsaved. See OpenAI’s Library file management guide and Space getting started for where saved files land.

Temporary Chat, archive, delete, and training settings

Temporary Chat is easiest to understand when compared with the nearby controls. The following table describes their purposes rather than implying they offer equivalent privacy. Choose the control that matches the actual problem you want to solve.

Archiving is mainly an organization choice. A conversation remains in the account and can be found and restored through the archive controls. Read how archived chats work if your main concern is a crowded sidebar.

Deletion removes an existing conversation and is irreversible through the service’s normal recovery options. It does not automatically delete a separate saved memory or separately saved Library file. Follow the ChatGPT history deletion guide when you want to remove content already stored in the account.

The training control concerns eligible use of personal-account content for model improvement. It does not automatically turn ordinary chats into temporary ones. A chat can remain visible in your history even when you have changed the model-improvement preference.

OpenAI has a separate chat and file retention guide that explains retention periods for temporary chats, saved chats, and files in Library or Space. If you need to know exactly how long a specific type of content is kept, check that page rather than assuming “temporary” means “immediately deleted.”

Third-party actions and work accounts need another check

A GPT or connected tool can send information to an outside service as part of an action. OpenAI’s temporary-chat documentation says that the recipient’s privacy policy governs that data and that the recipient may retain it for longer than 30 days. Do not assume the temporary label follows information wherever it travels.

Before allowing an action, inspect which service receives the data and what the action will do. A draft inside a chat differs from an email sent to a real person or an event created in a real calendar. Review recipients, content, and permissions before approving an external change.

Workspaces can impose additional restrictions and retention requirements. A personal privacy preference cannot guarantee that an employer’s account behaves like a personal account. Use approved tools and ask the workspace administrator if the available options are unclear.

For shared devices, also consider local traces. A screenshot, clipboard entry, download, or copied document can outlast a temporary conversation. Sign out when finished and avoid leaving private material open on a device that other people use.

A five-minute exercise to learn the mode

Use this exercise to recognize the controls without sharing personal information. The fictional task deliberately does not require memory, files, connected apps, or current factual research. You can focus on the mode indicator and the choice to preserve the result.

1. Start a new temporary chat and choose Unpersonalized if offered.
2. Ask: “Suggest four friendly names for a fictional neighborhood puzzle club. Use fewer than six words per name.”
3. Read the result and ask for one revision, such as a calmer tone.
4. Locate the temporary indicator again and identify any save option without activating it unintentionally.
5. Decide whether you need the names. Copy one harmless name into a practice note if you want to keep it.
6. Leave the conversation and review what is, and is not, available in normal history.

The expected learning outcome is knowing why the conversation is temporary and what saving would change. A missing sidebar item does not measure the service’s safety-retention period. Avoid using a test phrase containing a real secret to probe later personalization.

For a second practice session, choose Personalized with another harmless prompt and observe the mode choice. You do not need the assistant to reveal private memories to prove that you understand the distinction. If you want to inspect remembered information, use the dedicated memory controls instead.

Common problems and what to do next

If you cannot find Temporary, check whether you are starting a new chat and whether your workspace restricts the feature. Review the official guide and the app’s current controls. A rollout difference is more plausible than assuming that every old screenshot must match your account.

If a useful chat has disappeared, check whether it was temporary and unsaved. An ordinary archive search cannot restore an unsaved temporary conversation that was never added to history. Keep useful information intentionally while the conversation is available.

If you accidentally saved a temporary conversation, review it as an ordinary chat under your account settings. You can decide whether to delete the saved chat and inspect relevant memory sources. Saving and later deleting does not justify a claim that the conversation was never processed under ordinary settings.

If you pasted information you should not have shared, stop sending more of it. Review account controls, any connected recipients, and the appropriate incident process for the information involved. Do not rely on a chatbot’s reassuring message as confirmation that every copy has been removed.

Questions beginners often ask

Can Temporary Chat use my existing memories?

Yes, when you choose Personalized under the current documented options. Choose Unpersonalized before the first message if you do not want existing memories, custom instructions, and plugins used. Temporary chats do not create or update memories while they remain temporary.

Can I save a temporary conversation?

Current OpenAI documentation allows saving it as an ordinary chat. Saving means normal account settings for personalization, model improvement, and retention apply. Review that change before choosing Save.

Can I recover an unsaved temporary chat tomorrow?

Unsaved temporary chats do not appear in ordinary history and cannot be reopened later through that history. Preserve a useful result intentionally while it is available. Keep in mind that copying text creates a separate copy wherever you put it.

Is Temporary Chat completely confidential?

No. OpenAI may keep a copy for up to 30 days for safety, and outside services have their own policies for data sent through actions. Workspace restrictions and limited safety-context uses can also apply.

Can my employer see my temporary chats?

If you are in a workplace or school workspace, temporary chats are available to authorized compliance tools for 30 days through the Compliance API for eligible Enterprise customers. Workspace administrators may have different retention settings. A personal-account temporary chat does not have this compliance layer. Ask your workspace administrator if you need to know what applies to your account.

When to ask a person

Ask an administrator or security contact before using workplace, customer, medical, financial, or other restricted information. Contact official support for persistent control problems, and ask a qualified professional to handle decisions where a mistaken answer would affect someone’s rights, health, or money.

Keep learning in small steps

Review ChatGPT history deletion for existing conversations and the beginner AI privacy guide for upload decisions. You can also take the AI level test to find a practical safety habit to work on next.

A final reminder

Choose a chat mode deliberately, then share only the information the task needs.