ChatGPT Temporary Chat Explained: The New Personalize and Save Controls

Temporary chat is the ChatGPT mode that leaves no trace: it stays out of your chat history and, by default, ignores everything ChatGPT knows about you. On August 27, 2026, OpenAI changed what it can do. You can now choose to have a temporary chat use your memories, custom instructions and plugins, and you can choose to keep one afterwards. The second choice is the one to read carefully, because saving a temporary chat ends the thing that made it temporary.

Key Takeaways

• Nothing changed unless you opt in. OpenAI says temporary chats are "not personalized" by default, and a non-personalized one still uses no memory, no custom instructions and no plugins.
• You can now personalize one. A personalized temporary chat "can use your memories, custom instructions, and plugins to tailor responses."
• A personalized temporary chat still does not remember you. OpenAI says personalized temporary chats "do not create new memories and remain out of your chat history unless you save them."
• The choice is locked at the start. Personalization "can be set only when you start a temporary chat and cannot be changed after the conversation begins."
• Saving one converts it into a regular chat. OpenAI's words: it "then follows your account-level personalization settings and model-improvement preferences." Model-improvement preference is the training setting, so saving hands the conversation to whatever that setting says.
• OpenAI names no plan, no country and no device. The release note says only that the controls are "rolling out."

What OpenAI Actually Changed on August 27

Two controls, both optional, both attached to temporary chat. OpenAI's release note is titled "More controls in temporary chat" and says: "You can now choose to personalize responses in a temporary chat with memory, plugins, and custom instructions from your regular chat settings. You can also choose to save a temporary chat."

That is the whole announcement. It does not change the default, it does not change what a normal chat does, and it does not touch your existing history. If you have never opened a temporary chat, nothing on this page applies to you yet.

Personalized vs Non-Personalized: The Actual Difference

The two modes differ on exactly one axis: whether the chat is allowed to read what ChatGPT already knows about you. Neither one writes anything back.

The useful way to think about it: a personalized temporary chat borrows your ChatGPT profile without adding to it. It is the setting for "answer me like you know me, but do not file this away" -- a question about a medical symptom, a salary, a relationship, a job you have not told anyone you are applying for.

The lock is the part people will trip over. You pick when you open the chat, and that is final for that conversation. If you started a plain temporary chat and then wish ChatGPT knew your usual context, there is no toggle -- you close it and open a new one.

The Part Worth Reading Twice: Saving

Saving a temporary chat is not "keep a copy." It is a conversion, and OpenAI says so plainly: "Saving converts it into a regular chat, so it then follows your account-level personalization settings and model-improvement preferences."

Unpack that second half. Model-improvement preference is OpenAI's name for the account setting that decides whether your conversations can be used to improve its models. A temporary chat sits outside that system while it is temporary. The moment you save it, it becomes an ordinary chat in your history and is governed by whatever that account setting currently says. OpenAI's own example is about the other half of the same rule: "if you have memory enabled and save a temporary chat, this will become part of your chat history and can be referenced to personalized future responses."

So the honest one-line version: if you started a temporary chat because you did not want the contents in your account, do not save it. If you do want to keep it, check your model-improvement setting first, because saving applies that setting retroactively to a conversation you had under different rules.

Who Gets It, and When

OpenAI has not said, and the gap is worth stating rather than guessing. The August 27 release note names no plan, no country, no age gate and no platform. It says the controls are "rolling out," which means the change is in progress rather than finished, so the absence of the option in your account today is not evidence that you are excluded.

We will not fill those blanks from third-party write-ups. When OpenAI publishes a plan or region gate, this page gets a dated update.

How To Use Temporary Chat At All

Temporary chat is the mode OpenAI already points people to when they want a conversation that does not feed the system. Its memory FAQ says it twice: "If you don't want information from a chat to be used to personalize responses, you can turn off memory or use Temporary Chats," and "To chat without using saved memory, use Temporary Chat."

The blunter instrument is still there if you want it. The same FAQ says you disable saved memories entirely by turning off Enable memory in Settings > Memory, and that you can delete individual memories or clear all of them. Temporary chat is the per-conversation version of that; the settings toggle is the permanent one.

Should You Use the New Controls?

Personalize a temporary chat when the context helps and the record does not. Asking about a health worry, a money problem, or a job move where you want ChatGPT to already know your situation but you do not want the conversation stored -- that is exactly the case this was built for.

Leave it non-personalized when you are testing what ChatGPT says without your profile shaping it. If you want a clean answer that is not shaped by everything it has learned about you, the default is still the right mode, and it is still the default.

Do not save a temporary chat you would not have started as a normal one. This is the only genuine trap in the release. Saving is presented as a small convenience and it is actually a change of governing rules for the whole conversation.

Do nothing if you have never used temporary chat. Your normal conversations are unaffected, the default has not moved, and there is no setting you now need to go and check.

For the other side of this -- what ChatGPT keeps on purpose and how to control it -- see our explainer on ChatGPT project memory, and Claude's memory for how Anthropic handles the same question. Our page on ads in ChatGPT covers what OpenAI says about where ads can and cannot appear, temporary chats included.

The bottom line: two opt-in controls, one default unchanged, and one sharp edge. Personalizing a temporary chat is safe by OpenAI's own description -- it reads but does not write. Saving one is the opposite: it takes a conversation you deliberately kept outside your account and files it inside, under your account's rules.

Sources

• OpenAI, "ChatGPT Release Notes", entry dated August 27, 2026: "More controls in temporary chat."
• OpenAI, "Memory FAQ", on temporary chats, saved memories, and the Settings > Memory switch.