Copilot Memory Settings: Review, Enable, or Turn Memory Off

Copilot memory settings depend on the account and app: the current consumer app uses Settings, Personalization, while Microsoft 365 Copilot Chat provides its own Chat settings, Personalization controls. Review the specific choices before enabling or disabling memory, because saved entries and other personalization sources are separate. fenbs keeps intentionally chosen project context outside these personal settings for supported connected assistants.

How do you find the right memory controls?

1. Confirm the app and signed-in account. Distinguish a personal Microsoft account from your work or school account.
2. In the current consumer app, open Settings, then Personalization, and review Saved memories.
3. For Microsoft 365 Copilot Chat, open Settings and more, Chat settings, then Personalization.
4. Set Saved memories to On to enable saving, or Off to disable future saving. Review the other personalization controls separately for their own effects.

Microsoft’s privacy controls for individuals↗ (opens in a new tab) document the consumer setting. Microsoft’s saved-memory guide↗ (opens in a new tab) documents the Copilot Chat path. Use the guidance that matches your account instead of combining steps from different screenshots.

A useful starting choice is a narrow one: “I want it to remember my response preference,” or “I want a conversation that does not reuse personal context.” Those are clearer goals than “turn everything off.” Once you know the outcome, you can choose and verify the corresponding control.

Which settings do different jobs?

Think of settings as controls over particular information, rather than one master switch for everything an assistant can know. A record you saved, a conclusion drawn from old chats, and a document you attach to the current message are different sources. Changing one source does not explain the others.

Microsoft’s personalization overview↗ (opens in a new tab) explains saved memories, chat-derived information, eligible work insights, and custom instructions as different inputs. Work insights depend on the work or school account and its available experience. Check the current vendor guidance for availability instead of assuming a colleague’s menu will match yours.

For example, if you wrote “always answer in bullet points” in custom instructions, turning off future memory saving does not address that instruction. Review the record that contains the preference you want to change. For work choices a teammate must inspect, fenbs keeps the explicit project record beside its tasks.

What changes when you enable or disable saved memories?

In the documented controls, Saved memories governs the ability to save memories. Turning it off does not automatically delete entries already saved. If removal is your goal, use the management list and the dedicated clear Copilot memory guide rather than treating a switch as a deletion confirmation.

Before enabling it, choose a harmless preference you would actually find useful, such as a preference for a brief summary before details. Check the saved record and later response. You do not need to fill memory with personal facts to decide whether a small amount of personalization helps your work.

Before disabling it, consider which preferences you still need to provide in a task-specific request. “Summarize this report in five sentences” is a clear current instruction whether or not you use persistent personalization. A precise prompt can reduce the need for a permanent preference.

Keep model choice separate from memory control. The Copilot and ChatGPT comparison addresses assistant selection. A memory setting is a choice about supplied context, not a guarantee that every answer will follow your preference perfectly.

How do work chat history and temporary chat affect the choice?

Microsoft’s Copilot Chat history guidance↗ (opens in a new tab) describes a separate Chat history personalization toggle. Some work accounts label it Chat history & work insights. Turning it off starts removal of the relevant inferences; re-enabling it during Microsoft’s documented deletion period can restore them. Read that explanation before using the toggle as a reset.

For a single conversation without using or updating personalized information, the work personalization guidance describes temporary chat. It also explains that an organization’s retention policy can still apply. Choose temporary chat for the conversation boundary it provides, and ask your organization about retention when that is your actual concern.

Do not describe turning off personalization as deleting company documents, changing access rights, or erasing every work record. Those are different actions. If the unwanted context comes from a file you attached in this conversation, start by reviewing that attachment and the current request.

Why might your settings look different?

First rule out the wrong account or product. A browser can hold both a personal and a work identity. Check the identity shown in Copilot itself before comparing screens. Note the precise label of the missing option, the app, and the account type so an administrator or support person has something concrete to investigate.

GitHub Copilot is a separate development product, as the GitHub Copilot app guide explains. Its coding preferences and repository memory are not managed by the consumer Microsoft Copilot switch. Use GitHub Copilot memory for that neighboring question instead of trying to make these account settings control an editor.

Avoid repeatedly toggling nearby controls while looking for one that seems to work. That changes several variables at once. Make one intended change, inspect the relevant record, and test a small new conversation. Our explanation of agent memory helps separate a stored record from information in the current prompt.

How to manage selected project context with fenbs

1. Create a free account, free for one person with no credit card, and add one active project.
2. Add its current tasks and the approved decisions that affect the next action. Keep personal preferences out of the shared project record unless they are actually relevant.
3. Connect a supported assistant through Integrations. MCP is a standard way for an AI app to use an outside tool; fenbs connects through browser sign-in.
4. Ask the assistant to read the project context before work and save task progress plus Where we left off before it stops. Review the saved summary yourself.

fenbs is not the right fit for controlling Microsoft’s memory switches. It does not connect to consumer Microsoft Copilot, read its chat history, or import its memories. Use it for the work you intentionally retain and share through supported assistants.

Review your setup after a real task. Did personalization save editing time? Could the next person find the approved choice and remaining work? These are separate tests of usefulness. The AI memory hub explains the wider options for keeping context accessible and correctable.