AI/TLDR

Anthropic · 2026-08-25 · major

Claude memory works everywhere — one memory across chat and Cowork

Claude's memory now spans chat and Claude Cowork, so context saved in one shows up in the other. Claude writes topics during a conversation instead of summarizing it afterward, and every topic is a file you can read, edit, or delete.

Anthropic announcement graphic for Claude memory across chat and Cowork

Claude and Claude Cowork now draw on the same memory, and every saved topic is a file you can open and change.

Quick facts

MakerAnthropic
What's newOne memory shared by Claude chat and Cowork
Where it worksWeb, desktop, and mobile
On by defaultFree, Pro, and Max plans
Team and EnterpriseOff unless an admin turns it on
Sensitive topicsExcluded by default, opt-in setting
Never storedSSNs, government IDs, criminal history, immigration status

What is it?

Claude's memory now works across chat and Claude Cowork rather than staying inside each one. A cloud task running in Cowork can use what you told Claude in chat, and anything Cowork learns comes back to chat. Memory is on by default on the Free, Pro, and Max plans, across web, desktop, and mobile.

How does it work?

Topics are written during the conversation instead of being summarized once it ends. Each topic Claude keeps is stored as a file, and the full list sits under Topics in Memory settings, where you can read it, correct it, or delete it. Fixing one file carries forward: correct an old company name once and every conversation after that uses the new one.

Why does it matter?

Re-explaining your project, your stack, or your writing style in every new session is the tax people pay for chat assistants, and sharing one memory between chat and Cowork removes it. The editable topic list also answers the usual objection to memory features. You can see exactly what was kept about you and change it, instead of trusting a summary you never get to read.

Who is it for?

Claude and Cowork users

Frequently asked questions

Which Claude plans have memory turned on by default?
Memory is on by default for Claude Free, Pro, and Max users on web, desktop, and mobile. Team and Enterprise organizations work the other way around: an administrator decides whether memory is available, and individual users then switch it on themselves. TechCrunch notes that iOS and Android users need the latest app version for the update.
Does Claude store health or religious information in memory?
Claude leaves sensitive topics out of memory by default, including health, beliefs, race, ethnicity, religion, politics, and gender identity. Turning on the sensitive-topics setting allows them, and Claude then shows a notice each time it saves something in that group. Social security numbers, government IDs, criminal history, and immigration status are never stored either way.
Can I keep Claude chat and Cowork memories apart?
Not inside one account. The Register reports there is no way to split Claude chat memory from Cowork memory, and lists the workarounds: separate accounts, pausing memory, incognito chats, or deleting individual topics by hand. The same report says Claude Code keeps its own memory and was not folded into this change.
How do I remove something Claude remembered about me?
Everything Claude has kept appears as a file under Topics in Settings > Memory, where each entry can be read, edited, or deleted. Because a topic is one file used by both chat and Cowork, a single correction applies in both places, so an outdated employer or project name only has to be fixed once.

Try it

Open Settings > Memory in Claude to see the Topics list.

Sources · 4 outlets

Tags

  • anthropic
  • claude
  • claude-cowork
  • memory
  • personalization
  • privacy
  • assistant
  • release

← All releases · Learn AI