Claude's Memory Updates in Real Time Across Chat and Cowork

Claude’s memory updates live. Claude now saves facts to memory as you type them, instead of waiting to summarize the chat history.

You can read and edit exactly what it knows. Memory now shows up as a list of short, separate files by topic (in Settings > Memory), rather than one long summary. Whenever you correct something, like your company’s old name, it’s fixed everywhere Claude uses memory.

For example, when I looked at my Claude memory, I noticed that something I only ran as a test was included. By clicking on “Edit” and prompting Claude to remove it, I was able to remove it.

Sensitive topics stay off by default. Health, race, ethnicity, religion, politics, and gender identity are excluded automatically, but you can opt in with a toggle.

Security information like government IDs or Social Security numbers stays out of memory regardless of your settings.

The new Memory is available for Free, Pro, and Max plans across web, desktop, and mobile.

For Team and Enterprise, an admin must turn availability on for the organization first, and individual employees have to opt in.

If you’re on Team or Enterprise, confirm with your admin whether memory is switched on.

By toggling Memory logs under Settings, you can monitor and control what your AI remembers and who has access to what information.

For example, when I took a look, it had a few details in there that I wouldn’t want Claude to have as context in every chat.

Since memory now lives as separate and readable files, you can exclude sensitive topics and correct what’s wrong anytime.

This protects you and your organization while giving Claude exactly what it needs to help you.