Anthropic announced an update to Claude Cowork's and chat function's memory mechanism, integrating the two systems' memories and managing them uniformly. The official stated that Cowork now has memory and can reuse the same memory in chats, so users no longer need to repeatedly explain the background, and Claude can better continue previous work.
Sensitive information is not stored by default and can be manually enabled
The logic of bidirectional integration is: when Cowork performs tasks in the cloud, it can read the information remembered in chats, and the new memories generated by Cowork can also be used in chats. Anthropic emphasized that memories related to sensitive topics such as health and personal beliefs are disabled by default, and users can manually enable them in the "Memory" settings; free and paid versions, Mac and iOS are supported.
According to the official statement, the memory function is enabled by default for the Free, Pro, Max packages on web, desktop, and mobile platforms, while sensitive items remain disabled by default. iOS and Android users need to update to the latest version. For heavy users, this means turning "personal context" into a persistent asset across workstations, but also returning the switch of "which information is long-term remembered" to the user's hands.
