Why won’t Claude remember my personal life or health?
Claude’s memory does keep personal things. As of November 9, 2026, Anthropic’s help article lists “the people and places in your work and life” among what Claude remembers. What it leaves out by default is narrower: “personal or sensitive subject matter, like your health, race, ethnicity, religious beliefs, politics, gender identity, and other similar areas.”
You can change that. Turn on Include sensitive topics in memory in Settings > Your Claude, then tell Claude again, since nothing said before you turned it on is saved after the fact. A short list is never saved, even if you ask: government ID numbers, criminal history, financial account numbers and immigration status.
If what went missing came from an import, such as your ChatGPT memory, that’s a separate filter: Anthropic says imported “personal details unrelated to work” may not be kept. Add them back by hand, as shown below.
This page is a research summary of Anthropic’s help pages and announcement, read on November 9, 2026. An AI drafted this page from the sources below and another AI checked it. No person has run these steps. Counterparts is ours: Mike LaPeter runs Counterparts.
What Claude’s memory keeps, and what it leaves out
From Anthropic’s memory article, read November 9, 2026. Anthropic changes these rules, so check the article if something here doesn’t match what you see.
| Kind of detail | Saved by default? | With Include sensitive topics on | How to change it |
|---|---|---|---|
| Everyday context: your role and projects, the people and places in your work and life, how you like to communicate and work | Yes, as Claude picks it up while you chat | Yes | Say “remember this”, or edit it under Topics |
| Health, race, ethnicity, religious beliefs, politics, gender identity “and other similar areas” | No | Yes, from when you turn it on, with a notice each time | The setting in Settings > Your Claude |
| Government ID numbers, criminal history, financial account numbers, immigration status | Never, “even if you ask” | Never | Keep these in your own notes |
| Anything from an incognito chat, or a chat started with Memory off in the + menu | No | No | Start a normal chat |
| A chat where Claude used your linked accounts in Personal Finance | No: “it’s designed not to save anything from that chat to memory” | No | By design, no setting |
| Personal details in a memory you imported from another AI | May not be kept, if they’re “unrelated to work” | Not stated; see below | Add them in Manage or in a chat |
Anthropic’s August 25 announcement words the never-saved list a little differently: “sensitive identification numbers (SSN, government ID numbers, etc), criminal history, immigration status, or anything that violates our Acceptable Use Policy.” It gives one example of a sensitive topic someone might want kept: with the setting on, “Claude will remember things like your gluten allergy when suggesting recipes for weekly meal prep.”
Turning on sensitive topics
- Go to Settings > Your Claude (in the older layout, Settings > Memory) and turn on Include sensitive topics in memory. Claude also offers this in a one-time notice in the chat, the first time it declines to save something because it touched a sensitive topic.
- Tell Claude the detail again. Anthropic says “Anything from before you turned it on isn’t saved retroactively.”
- Watch for the notice. Each time Claude saves something on one of these topics, a notice appears above the message box so you can review it or change your settings.
- Check Settings > Your Claude, Manage next to Your memory, under Topics. That’s where Anthropic says everything Claude remembers is listed.
On the Claude app for iPhone or Android, update to the latest version first. Anthropic says that on an older version you won’t see the notice, “and Claude won’t save the sensitive topic to its memory.”
Two things remove what it saved. If you decline the notice, or turn the setting off later, “Claude removes any sensitive items already saved to memory.” Pausing memory keeps them. Anthropic says “If you unpause memory, both your memory and sensitive topics memory will be on,” so check the setting after you unpause.
On a Team or Enterprise plan, memory and sensitive topics are separate controls. An owner decides whether sensitive topics are allowed, and even then “nothing in those categories is saved until each user opts in themselves.” Anthropic says owners “can’t view or edit a user’s individual memories”, but its article also says memory entries are included in standard conversation history exports. Organizations with HIPAA or public-sector agreements don’t get the memory this page describes.
Why a personal detail still didn’t stick
If the detail isn’t on the sensitive list, Anthropic’s articles leave three explanations.
Claude didn’t pick it up. Claude saves topics on its own as you chat, and its articles don’t say how it chooses. Say “remember this” with the detail, then check Topics. If it isn’t there, add it with the “Tell Claude what to change or remove” box in Manage.
You’re on the older memory. A few Team and Enterprise organizations still have it. If you see Memory under Settings > Capabilities, that’s you, and Anthropic says the older memory “focuses on work-related context.” Its list leaves out “the people and places in your work and life.”
It came from an import. See the next section.
If the problem is Claude forgetting in general, not only personal things, Claude forgets between chats goes through every fix.
Imported memory
Anthropic’s import tool takes a memory export from another AI, such as ChatGPT, and Claude turns it into entries of its own. Its article says Claude’s memory “is designed to focus on work-related topics to enhance its effectiveness as a collaborator. As a result, Claude may not retain imported personal details unrelated to work.” It also calls imports experimental: “Claude may not always successfully incorporate imported memories.”
To get a missing detail back, go to Settings > Your Claude, select Manage next to Your memory, and type it into the “Tell Claude what to change or remove” box, or ask Claude in a chat.
The import article doesn’t say whether the sensitive-topics setting or the never-saved list applies to imports. Claude stores an import “as individual memory entries”, so expect the same rules as for anything else in memory, though we haven’t tested it.
Anthropic’s export prompt asks the other AI for “Personal details: name, location, job, family, interests”, and its article says you can edit that prompt “to exclude any sensitive information before importing into Claude.”
See and edit what Claude remembers
Settings > Your Claude, then Manage next to Your memory. Everything is listed under Topics: open one to read it, use the edit icon to change it, or Delete to remove it. A fix “applies to every conversation from then on.”
You can also tell Claude in a chat what to remember, change or forget. To see it all at once, Anthropic’s import and export article gives this prompt: “Write out your memories of me verbatim, exactly as they appear in your memory.”
Deleting a chat doesn’t delete what Claude remembered from it. Anthropic says “memory entries generated from it aren’t removed”, so delete those under Topics. How to make an AI forget something covers this for other assistants too.
Places you write yourself
Two places hold only what you put there, so Claude doesn’t choose what goes in.
- Instructions, in Settings > Your Claude. Anthropic says “Any instructions you add here will be applied to all of your conversations with Claude.”
- A project, at claude.ai/projects. Files you add to the project knowledge are “used across all of your chats within that project”, and project instructions apply to chats in that project. Free accounts can make up to five projects. Project instructions may need a paid plan.
Anthropic’s articles describe the sensitive-topics rule for memory only, and don’t say how it applies to Instructions or project files. Whatever you put there is sent to Anthropic with every chat it applies to.
Before Claude keeps health details
The setting decides what’s kept between chats. Anything you type in a chat is already sent to Anthropic, whether or not it’s saved to memory. Here’s what Anthropic’s privacy pages, read November 9, 2026, say about who can read it.
- Anthropic staff: “By default, Anthropic employees cannot access your conversations unless” you share them as feedback, or “Review is needed to enforce our Usage Policy”, in which case only “designated members of our Trust & Safety team” may access them.
- Training, on Free, Pro and Max: Anthropic uses your chats to improve its models if you allow it in your settings, or if a conversation is “flagged for safety review”. Incognito chats aren’t used, “even if you have enabled Model Improvement”.
- Work accounts: “Memory entries are included in standard conversation history exports”, and an organization’s data retention period can apply to memory too.
Who can read what your AI’s memory keeps goes through this for Claude, ChatGPT, Gemini, Copilot and others.
A saved condition is only as accurate as what was written down. Open it under Topics and correct it if it’s wrong. Claude’s memory isn’t a medical record, and a saved detail doesn’t make Claude’s advice on it any more reliable.
Counterparts: the memory we make
Claude’s own memory saves topics about you. Counterparts gives Claude one memory of the whole relationship: who you are, how you like to work, what you decided together, what’s in progress and where you left off. Claude writes it down as notes in its own words, and keeps a journal of your time together. What matters sticks, and the rest fades unless it comes up again.
That includes your life outside work. Counterparts has no work-only focus and no setting to opt into, so when you mention a family matter, a belief or a condition, Claude can write it down and bring it up next time without you explaining it again. It decides what’s worth keeping, so a health detail you mention may be kept whether or not you meant it to be. The local install takes credentials in recognizable shapes, such as API keys and tokens, out of what it writes down. It also catches most passwords named in a sentence, locally and on hosted. It has no rule for health, beliefs or ID numbers. Tell it only what you’d be comfortable having kept.
Claude Code on a Mac or Linux. (On Windows, install with npm and Node 22.15+ instead, from the Quickstart's On Windows section; it's new there, so tell us how it goes.) Run these in Claude Code, then restart it:
/plugin marketplace add mlapeter/counterparts
/plugin install counterparts@counterpartsIt’s free and open source, with no account or API key, and uses your own Claude plan. The memory is a file on your machine, readable by your account, and the conversation it writes up from stays there too: about three days after it’s written up, or about two weeks after the conversation goes quiet. To see what it has written down, type /counterparts:dashboard in Claude Code. It writes in its own words, so check what it kept about anything that matters, such as a condition. The Quickstart’s terminal install has a counterparts remove command for deleting a memory. What Claude reads from it, in a session or in a once-a-day background run, goes to Anthropic like the rest of your Claude Code conversation. Counterparts for Claude Code has more.
Claude Desktop on a Mac. Follow the Quickstart’s Install section, quit Claude Desktop, then run counterparts install --host claude-desktop --name "Your Name" --budget 9000 with your own name. Reopen Desktop and set the counterparts tools to Always allow. It’s free too. The memory stays on your Mac, apart from what a chat reads from it.
claude.ai, on the web and in the apps. A hosted version is coming soon, in beta, with nothing to install. Sign up on the home page to hear when it opens, or read how it will work. Its memory will be kept on a server we run, one store per person, and it isn’t end-to-end encrypted, so we could technically read it. “We don’t read it” is our promise, not something the system enforces or records. What you say in claude.ai goes to Anthropic as usual. How hosted handles health details hasn’t been tested, so weigh that before you share them there.
Questions people ask
Why does Claude’s memory only save work stuff?
For most people it doesn’t. Anthropic’s memory article lists “the people and places in your work and life” among what Claude remembers. Three things leave personal details out: sensitive topics such as health, religion and politics aren’t stored unless you turn on Include sensitive topics in memory; the import tool “may not retain imported personal details unrelated to work”; and a few Team and Enterprise organizations still have the older memory, which “focuses on work-related context”. For an ordinary personal detail, say “remember this” and check that it shows under Topics.
Why won’t Claude remember my health condition?
Health is one of the topics Claude leaves out of memory by default, with race, ethnicity, religious beliefs, politics, gender identity “and other similar areas”. To change that, turn on Include sensitive topics in memory in Settings > Your Claude, then tell Claude again: it saves these topics going forward only. On a phone, update the Claude app first, because an older version won’t save them. Before you do, read what to weigh below.
Where is the setting to include sensitive topics in Claude’s memory?
Settings > Your Claude (Settings > Memory in the older layout), under Include sensitive topics in memory. Claude also offers it in a one-time notice in the chat, the first time it declines to save something because it touched a sensitive topic. On Team and Enterprise plans, an owner has to allow sensitive topics for the organization, and then each person turns it on for themselves.
What will Claude never save to memory?
Anthropic’s memory article says some information “is never saved to memory, even if you ask”: government ID numbers, criminal history, financial account numbers and immigration status. Anthropic’s August announcement adds Social Security numbers and anything that violates its Acceptable Use Policy. The sensitive-topics setting doesn’t change this list, and Claude tells you when it can’t save something for this reason.
Does Claude’s memory leave out religion and politics?
By default, yes. Religious beliefs and politics are on Anthropic’s list of sensitive topics that stay out of memory until you turn on Include sensitive topics in memory. Anything you said about them before you turned it on isn’t saved after the fact, so say it again.
I imported my ChatGPT memory and my personal details disappeared. Why?
Anthropic’s import article says Claude’s memory “is designed to focus on work-related topics” and “may not retain imported personal details unrelated to work”. It also calls imports experimental and says Claude “may not always successfully incorporate imported memories”. To add a missing detail, go to Settings > Your Claude, select Manage next to Your memory, and type it into the “Tell Claude what to change or remove” box, or ask Claude in a chat.
Should I turn on sensitive topics in Claude’s memory?
That’s your call, and it comes down to what you want kept and who can read it. With it on, Claude saves health and similar details from then on, shows a notice each time, and you can review or delete each one under Topics. Anthropic says turning it off later removes them; for a work account its article says only that it stops saving new ones, so delete those under Topics yourself. Either way, what you say in a chat already reaches Anthropic, so the setting changes what’s kept between chats. Who can read AI memory sets out Anthropic’s staff access, training and work-account rules.
How do I see what Claude remembers about me?
Go to Settings > Your Claude and select Manage next to Your memory. Everything is listed under Topics, where you can read, edit or delete each one. You can also ask in a chat: “Write out your memories of me verbatim, exactly as they appear in your memory.” That’s the prompt from Anthropic’s import and export article.
Is there a personal memory for Claude outside work?
Claude’s own memory covers your life as well as your work, apart from the sensitive topics above. Counterparts, the memory we make, keeps one memory of you, your life and your work together, in Claude’s own words. It has nothing to opt into: personal and health details can go into memory like anything else. It’s free on Claude Code (Mac or Linux, and Windows, new, through npm) and Claude Desktop (Mac); for claude.ai a hosted version is coming soon, in beta, and you can sign up on the home page to hear when it opens. Claude chooses what to write down, so read its section before you tell it anything private.
Questions? Ask in r/AIMemorySystems.
References
- Claude Help Center: Use Claude’s chat search and memory, Import and export your memory, Release notes, Personalization features, Create and manage projects
- Claude: Claude’s memory works everywhere, and you decide what’s in it (August 25, 2026)
- Anthropic Privacy Center: How does Anthropic protect the personal data of Claude users?, Is my data used for model training?
- Counterparts: README, Quickstart, Counterparts for claude.ai, Privacy
- Related guides: Claude forgets between chats, Who can read AI memory, How to make an AI forget something