Claude’s Memory Now Crosses Chat and Work Sessions
Anthropic unifies Claude’s memory across chats and Cowork, giving users control over what it remembers.
The update
Claude’s long-term memory now persists when switching between the chatbot and the Cowork agent. Previously, context from one session did not automatically carry over to the other. This change allows users to maintain continuity across tools. Users can now see saved memories organized by topic, edit specific topics, and delete them from the settings menu. A new toggle allows Claude to include sensitive topics in memory, such as health details or preferences, though certain types of information are excluded by design.
Why it matters
Unified memory reduces the need to repeat context when switching between tools. For users who rely on Claude for both conversation and task automation, this update streamlines workflows. The ability to control what is stored and how it is organized provides more agency over personal data. This feature is available on web, desktop, and mobile across free, Pro, and Max plans.
What to watch
Anthropic has not retroactively added past conversations to memory. Users should verify that sensitive topics are enabled only if they want Claude to store that information. The system will notify users when it cannot save certain types of data, such as government ID numbers or financial account details.
Sources
- cnet.com — memory unification across Claude and Cowork, sensitive topic toggle, mobile availability
- engadget.com — memory unification across Claude and Cowork, sensitive topic toggle, retroactive behavior