You chat normally
As you talk, ChatGPT may extract a durable fact ("I'm vegetarian," "I code in Python") and write it to your saved memories — or simply leave a trace in your chat history.
ChatGPT memory explained
How ChatGPT memory works.
ChatGPT can remember things about you across conversations — your name, your preferences, the projects you're working on. This independent guide explains how that memory actually works under the hood, the difference between saved memories and chat history, and what it means for your privacy and your data — plus exactly how to view, control, or delete it.
Quick answer
ChatGPT remembers you in two ways
Memory isn't one thing. ChatGPT has a visible, editable list of facts, plus an invisible layer that learns from your past chats. They're controlled separately.
| Layer | What it is | Can you see / edit it? |
|---|---|---|
| Saved memories | Explicit facts ChatGPT stores about you — name, role, tone, preferences. Updated automatically or when you say "remember this." | Yes — a visible list you can view, edit and delete in Settings. |
| Reference chat history | Implicit recall of patterns from your past conversations (launched April 2025), used to make new chats more relevant. | No editable list — you can only toggle it on or off. |
| Temporary Chat | A session that ignores memory entirely and isn't saved to history. | Per-conversation — nothing is stored or reused. |
Both layers live in Settings → Personalization → Memory. Always confirm current behavior on OpenAI's Memory FAQ and Data Controls FAQ.
Under the hood
How memory actually reaches the model
The model isn't retrained for each user. Personalization happens at inference time: relevant memory is added to the context window with your prompt.
Memory is stored server-side
Saved facts and history are kept on OpenAI's servers, tied to your account — not on your device. Every prompt is processed server-side regardless of settings.
Relevant memory is retrieved
When you start or continue a chat, the system selects memory that's relevant to your prompt and injects it into the model's context, like a hidden system note.
The model conditions its reply
The model reads that added context alongside your message and answers as if it "remembers" you. Nothing about the base model's weights changed.
Go deeper: how memory works · saved vs chat history · memory vs custom instructions · memory limits
Privacy & data
What memory means for your data
| Question | What to know |
|---|---|
| Is memory used for training? | On consumer accounts, conversations (including memory) may be used to improve models by default. Opt out in Settings → Data Controls → "Improve the model for everyone." See memory & training data. |
| Does memory leave my device? | Yes — every message is processed on OpenAI's servers. Memory controls govern what's retained and reused, not whether data is sent. |
| How do I have a private chat? | Use Temporary Chat: no history, no memory read/write, not used for training; kept briefly (~30 days) then deleted. |
| Can I erase what it knows? | Delete saved memories individually or all at once, and turn off chat history. See delete memory. |
If memory worries you
Use AI without locking all your context to one provider
Memory is convenient, but it concentrates your personal context inside one company's servers. If you'd rather spread your usage — and pick which model sees which prompt — a multi-model workspace gives you that choice while you keep sensitive topics in memory-free sessions.
All-in-one
MultipleChat AI
multiplechat.aiUse ChatGPT, Claude, Gemini and other models in one place — choose where each prompt goes instead of feeding everything to a single assistant.
Privacy-first: MultipleChat doesn't save your chats to memory and doesn't share your data with model providers or let them train on it. Other privacy-focused options include Duck.ai, Proton Lumo, Brave Leo and Kagi Assistant; for full local control, self-hosted tools like Ollama or Jan keep everything on your device.
OpenAIChatGPT
chatgpt.comOffers Temporary Chat (no history, not used for training), a training opt-out under Settings → Data Controls, and editable memory. By default, consumer chats may be used to improve models unless you opt out — verify on OpenAI's help center.
AnthropicClaude
claude.aiAnthropic gives you data and training controls and has historically taken a privacy-conscious stance on consumer chats, but its training and retention settings have changed over time — check your current privacy settings in the app.
GoogleGemini
gemini.google.comGemini Apps Activity lets you pause history and delete past activity, but sampled conversations can be reviewed by humans, so avoid sharing confidential information. Review your Activity controls.
MicrosoftCopilot
copilot.microsoft.comMicrosoft 365 Copilot (business) doesn't use your organization's data to train models and adds Enterprise Data Protection; consumer Copilot has a privacy dashboard to review and delete your history. Check current Microsoft policies.
PerplexityPerplexity
perplexity.aiPerplexity has an "AI data retention" setting you can switch off so your searches aren't used to improve its models, plus account-level data controls. Confirm the current options in settings.
All guides
ChatGPT memory: every guide
How memory worksThe technical walkthrough.
Saved vs chat historyThe two layers compared.
Turn memory on/offManage the settings.
Delete memoryClear individual or all.
Memory & privacyRisks and how to reduce them.
Memory & trainingIs it used to train models?
Temporary ChatMemory-free conversations.
Memory not working?Troubleshooting steps.
Memory limitsHow much it can hold.
Team & EnterpriseMemory and data controls.
Memory vs custom instructionsWhat's the difference?
Memory FAQEvery common question.
FAQ
ChatGPT memory — quick answers
Short answers on how memory works, where it's stored, training, privacy and control.
How does ChatGPT memory work?
ChatGPT memory has two layers. Saved memories are explicit facts ChatGPT stores about you (name, job, preferences) that you can view and edit. Reference chat history lets ChatGPT draw on patterns from past conversations without an editable list. At the start of a chat, relevant memory is added to the model's context so answers feel personalized. Both are controlled in Settings → Personalization → Memory.
What is the difference between saved memories and reference chat history?
Saved memories are auditable: you can open the list, see each stored fact, edit or delete it. Reference chat history is not a visible list — it's implicit recall of earlier chats. Only the saved-memories list is observable. Turning off saved memories doesn't erase existing memories unless you delete them.
Where is ChatGPT memory stored?
On OpenAI's servers, tied to your account — not locally on your device. Every message is processed server-side regardless of memory settings; the controls determine what's retained and reused afterward, not whether data leaves your device.
How do I turn ChatGPT memory on or off?
Go to Settings → Personalization → Memory and toggle "Reference saved memories" and "Reference chat history" independently. Turning off saved memories stops ChatGPT using stored facts but doesn't auto-delete them. For a memory-free session, use Temporary Chat.
How do I delete ChatGPT memories?
In Settings → Personalization → Memory, open "Manage memories" to delete individual saved memories or clear them all. You can also tell ChatGPT to forget something in chat. Deletion and retention timelines are described on OpenAI's help center.
Is ChatGPT memory used to train OpenAI's models?
On consumer accounts, conversations (including memory) may be used to improve models by default; opt out in Settings → Data Controls → "Improve the model for everyone." Business, Team, Enterprise and API data are handled under different terms. Verify the current policy on OpenAI's site.
Does Temporary Chat use memory?
No. Temporary Chats don't appear in history, don't read or create memories, and aren't used to train models. OpenAI keeps a copy briefly (around 30 days) for safety, then deletes it.
Can ChatGPT memory be a privacy risk?
Memory stores personal details over time, so sensitive information can persist across chats and resurface later. Reduce the risk by reviewing saved memories regularly, using Temporary Chat for sensitive topics, and setting your data controls. Verify specifics on OpenAI's help center.
Can I see everything ChatGPT remembers about me?
You can see the full saved-memories list, but not the reference-chat-history layer, which has no visible list. To remove the implicit layer, turn off "Reference chat history" and consider deleting old conversations.
Does deleting a chat delete its memory?
Not necessarily. A fact saved to memory persists even if you delete the conversation it came from, so you may need to delete the memory separately. Deleting chats also reduces what the chat-history layer can draw on.
Do Team and Enterprise plans handle memory differently?
Business plans (Team, Enterprise) and the API generally offer stronger data protections, including not using business data to train models by default and admin-level controls. Memory availability and admin settings differ from consumer accounts. Check OpenAI's business documentation for current details.
How is memory added to a conversation technically?
Relevant saved memories and signals from past chats are injected into the model's context window alongside your prompt, like a system note, so the model conditions its reply on them. The model isn't retrained per user — personalization happens at inference time. Exact implementation is proprietary to OpenAI.
What's the safest way to use ChatGPT memory?
Decide what you want personalized, keep sensitive data (financial, medical, credentials) out of memory, review the saved-memories list periodically, use Temporary Chat for private tasks, and set data controls deliberately. If you use several AI tools, remember each stores context separately.