Anthropic has updated Claude’s memory feature so that the chat interface and Claude Cowork use the same context. Information learned in Chat can be transferred to Cowork, while information acquired during Cowork can be carried over to subsequent conversations. Claude can also now update its memory by topic while a conversation is ongoing, rather than only at the end of a conversation.
Users will be able to view, edit or delete the information Claude remembers under topic headings. Sensitive information such as health details, race, ethnicity, religious beliefs, political views and gender identity will not be saved by default. The “include sensitive topics in memory” option will need to be enabled for this information to be stored in memory. Government-issued identification numbers, Social Security numbers, criminal records and immigration status will not be stored even if this option is enabled.
The feature is enabled by default on the Free, Pro and Max plans and is available on the web, desktop, iOS and Android.
Why it matters
The update makes Claude more consistent across different use cases, reducing the need for users to repeat the same information in Chat and Cowork. This could provide a more cohesive experience for people who use the tool for both everyday conversations and work processes; however, it also makes it more important to keep track of what information the shared context contains. The options to view, edit and delete information by topic allow users to control this process. Excluding sensitive data by default and never storing certain identification information limits the scope of the memory feature. The remaining question is how consciously users will use the optional sensitive-topic setting.