ChatGPT Temporary Chat keeps a conversation out of your normal history, does not create or update memories, and is not used to improve OpenAI’s models while it remains temporary. OpenAI may retain a copy for up to 30 days for safety. A temporary chat can now be personalized or non-personalized, and saving it converts it into a regular chat with your account’s usual settings.
Last verified: 29 September 2026 against current OpenAI Help Center guidance. Controls vary by plan, app version and workspace policy.
Contents
- What Temporary Chat does
- Temporary versus regular chat
- How to start and configure it
- Other data controls
- What it cannot guarantee
- FAQs
What is ChatGPT Temporary Chat?
Temporary Chat is a conversation mode for situations where you do not want the chat to appear in normal history or create new memories. OpenAI says an unsaved temporary chat is not used to improve its models and may be kept for up to 30 days for safety purposes. That is a privacy control, not a promise of zero processing or instant deletion.
Before beginning, ChatGPT may let you choose whether the temporary chat is personalized. A non-personalized temporary chat does not use your existing memories, custom instructions or plugins for personalization. A personalized temporary chat can use those sources, subject to account and workspace settings, but still does not create or update memories while it remains temporary.
If you save a temporary chat, it becomes a regular chat. It then appears in history and follows the account’s personalization and model-improvement preferences. Saving is useful when a disposable conversation becomes worth keeping, but it changes the privacy behavior that motivated temporary mode.
Temporary Chat versus regular ChatGPT chat
| Question | Temporary Chat while unsaved | Regular chat |
|---|---|---|
| Appears in history? | No | Yes, until deleted or removed by policy |
| Creates or updates memory? | No | May, depending on settings and availability |
| Used to improve models? | No while temporary | Depends on account type and “Improve the model for everyone” |
| Retention | May be retained up to 30 days for safety | Saved until deleted or governed by workspace retention |
| Personalization | User can choose personalized or non-personalized where available | Uses account and workspace personalization settings |
The table describes OpenAI’s current product guidance, not a universal rule for every connected app. If you use plugins or apps in a personalized temporary chat, information sent to the connected service can be governed by that service’s own terms and retention. Review permissions before connecting sensitive data.
How to start a Temporary Chat
- Open ChatGPT on the official website or app.
- Start a new conversation and choose Temporary Chat, or open the Privacy Center option for a new temporary chat where available.
- Select personalized or non-personalized mode before sending the first message.
- Confirm the temporary indicator is visible.
- Avoid uploading information that the task does not require.
- When finished, close the conversation rather than saving it unless you intentionally want a regular chat.
Interface labels can change. If the option is missing, update the official app and check whether your plan or workspace administrator controls availability. Do not install a browser extension that promises to add “private ChatGPT mode”; use OpenAI’s own control.
Which Temporary Chat mode should you choose?
Choose non-personalized when
- You want the cleanest separation from saved memories and custom instructions.
- The task contains one-off information that should not influence later chats.
- You are testing how ChatGPT responds without personal context.
- You are using a shared workflow and want fewer personalization surprises.
Choose personalized when
- You want existing preferences or instructions to shape the reply.
- You understand that the chat may use memories or connected capabilities even though it will not create new memory while unsaved.
- The benefit of personalization outweighs the additional context exposure for the task.
You cannot assume “temporary” also means “anonymous.” A signed-in service can still process account, device and safety information. Use data minimization: summarize the situation, replace names with roles, remove identification numbers and share only the smallest relevant excerpt.
Example: getting help with a sensitive document
Anita wants help rewriting a complaint letter. She opens a non-personalized Temporary Chat and removes her name, address, account number, signatures and reference codes before pasting the text. She asks for clearer wording, copies the result and closes the chat without saving.
This is safer than uploading the full document, but it is not a guarantee of confidentiality. For legal, medical, employment or financial material, she should also consider whether the information may be shared with an AI service under the organization’s rules. Temporary Chat is a product control, not professional secrecy.
Turn off “Improve the model for everyone”
On eligible personal accounts, OpenAI’s Data controls include “Improve the model for everyone.” Turning it off means new conversations are not used to improve OpenAI models. OpenAI says the setting applies across devices when you are signed in. Turning it off does not remove saved chats from history and does not delete existing conversations.
- Open your account menu.
- Choose Settings, then Data controls.
- Turn off “Improve the model for everyone.”
- Review the setting after major account or app changes.
Business, Enterprise, Edu and ChatGPT for Healthcare workspaces are handled differently; OpenAI says their content is not used to train models by default. Workspace administrators may control available features, retention and export options. Ask the administrator when a control is unavailable rather than assuming the consumer instructions apply.
History, archive and deletion are different
Archiving a regular chat hides it from the main list but does not change its retention or model-improvement treatment. Deleting a saved chat removes it from your account view and schedules it for permanent deletion within 30 days, unless it has already been de-identified or must be retained for legal or security reasons. A deleted chat cannot normally be restored through support.
Files can have separate retention behavior. A file saved in a Library may remain after its chat is deleted; files added to projects or custom GPT knowledge can remain until the project, GPT or file is removed. Review each location rather than treating chat deletion as a universal file purge.
Memory controls and Temporary Chat
Memory and chat history are not the same. Deleting a chat does not automatically delete a memory saved from it. Review Settings → Personalization to inspect memory options and available sources. When you do not want a temporary conversation to use existing memories, choose non-personalized Temporary Chat before you begin.
OpenAI notes that limited safety-relevant context may still be used in rare, high-risk situations. This is separate from personalization. Therefore “non-personalized” means the conversation is not tailored using your selected memory and instruction sources; it does not disable safety systems.
Connected apps, plugins and external services
A personalized temporary chat can use plugins or connected apps when account and workspace settings allow. Data sent to an external service may be handled under that service’s privacy policy. Before using a connector, check the requested permissions, what content will be shared, whether the action writes to another service, and how to revoke access.
For broad product choices, our ChatGPT vs Gemini vs Copilot guide compares work-oriented trade-offs. For another provider’s controls, see the Gemini privacy settings guide. Features differ; do not carry one provider’s retention assumptions into another.
What Temporary Chat does not guarantee
- It does not make an unsafe prompt or illegal disclosure safe.
- It does not promise zero retention; OpenAI may keep a copy for up to 30 days for safety.
- It does not erase information already shared in another chat or external app.
- It does not delete files saved separately in Library, projects or GPT knowledge.
- It does not override an employer’s data-classification or confidentiality policy.
- It does not guarantee that every plan, region or workspace has identical controls.
Do not paste passwords, one-time codes, payment-card details, Aadhaar or PAN numbers, private medical records, confidential source material or unredacted customer data merely because the mode is temporary. If the task can be solved with placeholders, use placeholders.
Practical privacy checklist
- Choose Temporary Chat before entering sensitive context.
- Select non-personalized mode when you do not need memory, instructions or plugins.
- Redact names, account numbers, addresses and unique identifiers.
- Share excerpts rather than complete files.
- Review “Improve the model for everyone” in Data controls.
- Inspect Memory and connected-app permissions separately.
- Delete or remove saved files from every location where they remain.
- Follow workplace rules for confidential information.
Our ChatGPT features guide for Indian freelancers provides broader workflow context. This article is narrower: it answers what Temporary Chat and account-level privacy controls do, and where their limits begin.
Frequently asked questions
Are Temporary Chats used to train ChatGPT?
OpenAI says they are not used to improve models while they remain temporary. If you save one, it becomes a regular chat and follows your model-improvement preference.
How long can OpenAI keep a Temporary Chat?
OpenAI says a copy may be retained for up to 30 days for safety purposes.
Does Temporary Chat use Memory?
It does not create or update memories. Where available, a personalized temporary chat can use existing memories; choose non-personalized mode to avoid that use.
Can I recover a closed Temporary Chat?
An unsaved temporary chat is not in normal history. If you save it first, it converts to a regular chat. Do not rely on recovery; copy any non-sensitive result you need before closing.
Is turning off model improvement the same as Temporary Chat?
No. The account-level setting controls whether eligible new regular chats help improve models. Temporary Chat additionally stays out of normal history and does not create memories while unsaved.
Does deleting a chat delete its files?
Not always. Files saved separately in Library, projects or GPT knowledge may require separate removal.
Conclusion: temporary is a useful boundary, not invisibility
ChatGPT Temporary Chat is the right starting point for a one-off conversation you do not want in normal history or memory. Choose non-personalized mode when prior context is unnecessary, minimize what you share, and remember the up-to-30-day safety retention. Pair the feature with account-level data controls, memory review and careful handling of files and connected apps.
Official sources: OpenAI Temporary Chat FAQ; Data controls in ChatGPT; Chat and file retention.
Verification metadata: _cm_verified_source_url=https://help.openai.com/en/articles/8914046-temporary-chat-faq; _cm_source_authority=OpenAI Help Center; _cm_effective_date=Not applicable—current feature guidance.
Editorial metadata: Author: Ajit Naskar; Publication date: To be scheduled after approval; Updated: 29 September 2026; Last verified: 29 September 2026; Correction history: None.
Disclaimer: This independent guide is not affiliated with OpenAI. Feature availability, wording and workspace controls can change. Do not upload confidential data unless authorized.
Leave a Reply