Claude Memory Now Connects Chat and Cowork for a More Continuous AI Experience
Table of Contents
Anthropic has updated Claude memory to work across both its chat experience and Claude Cowork, allowing context built during conversations to carry into task-based workflows.
The change is designed to reduce the need for users to repeatedly explain project details, preferences and other information when switching between Claude chat and Cowork. Anthropic describes the update as a shared memory system, meaning information remembered in one experience can also be available in the other.
Claude Memory Now Works Across Chat and Cowork
With the new system, Claude memory is shared between chat and Cowork when Cowork runs tasks in the cloud.
For example, a user may have previously discussed how their manager prefers reports to be written. When asking Cowork to prepare an update, that previously established context can be available without requiring the user to explain it again.
Similarly, information about project priorities, team metrics or conference plans discussed in Claude chat can carry into relevant Cowork tasks. Information generated through Cowork can also become available in future chat interactions, creating a two-way memory system.
Memories Now Update During Conversations
Anthropic has also changed how Claude memory is created and updated.
Previously, the system relied on summarising conversations after they ended. Claude can now add and update memory topics while conversations are still taking place.
This means users may not need to explicitly ask Claude to remember every important change. For instance, if a project deadline changes during a conversation, that updated information can be reflected in future interactions.
Users can also pause or reset memory when they want greater control over what Claude retains.
Users Can Review, Edit and Delete Memories
The update gives users more direct control over their saved information.
Claude organises stored information under Topics in its Memory settings. Users can review individual topics and edit or delete information that has been saved.
This allows users to correct outdated information without having to repeatedly provide the same correction in future conversations or Cowork tasks.
Sensitive Topics Remain Off by Default
Anthropic has also introduced an optional setting for sensitive information.
By default, Claude memory does not store information relating to sensitive areas such as health, race, ethnicity, religious beliefs, politics and gender identity.
Users who want Claude to remember information from these categories can manually enable the Include sensitive topics in memory setting. When this option is enabled, Claude displays a notice when it saves information it considers sensitive.
The setting only applies going forward, meaning sensitive information from before the option was enabled is not automatically added to memory. Users can turn the setting off again at any time.
Some Information Is Never Added to Claude Memory
Even when sensitive-topic memory is enabled, Anthropic maintains restrictions on certain categories of information.
These include:
| Information Category | Claude Memory Status |
| Social Security and sensitive identification numbers | Not stored |
| Government ID numbers | Not stored |
| Criminal history | Not stored |
| Immigration status | Not stored |
| Information that violates Anthropic’s policies | Not stored |
These restrictions remain in place even when users choose to enable memory for selected sensitive topics.
Availability Across Claude Plans
Anthropic says memory is enabled by default for users on its Free, Pro and Max plans across supported web, desktop and mobile experiences.
The sensitive-topics option is turned off by default and must be enabled manually. For Team and Enterprise organisations, administrators control memory availability, with memory handled differently from the individual consumer plans.
The shared-memory update also does not extend to Claude Code, which continues to have a separate memory experience.
Conclusion
The latest Claude memory update connects Claude chat and Cowork more closely by allowing relevant context to move between conversations and cloud-based task workflows.
With real-time memory updates, editable topics and optional controls for sensitive information, Anthropic is giving users more continuity while also providing greater control over what the AI retains. However, the separation of Claude Code shows that the shared memory system currently remains focused on Claude’s chat and Cowork experiences.
