Anthropic announced a change to Claude’s memory system that unifies the memory used by chat and Claude Cowork, so information learned in one area is available in the other. The company said the update will stop users having to rebrief the assistant when switching between research and action-oriented tasks.
How the merged memory works
According to Anthropic, Claude will add topics to memory as conversations progress rather than waiting to summarize at the end. The company said that makes new memories available more quickly and shared between chat and Cowork even if a chat is still ongoing.
Anthropic provided an example in which Cowork would already recall details such as headcount, city and speakers for a conference because those items were discussed earlier in chat. The company said this should make it easier to move from developing ideas to drafting updates or agendas without re-entering prior details.
Controls, limits and notifications
Anthropic said users can read, edit or delete what Claude has stored in memory. By default, the system will not store certain personal or sensitive categories — including health data, race, ethnicity, religious beliefs, politics and gender identity — but users may enable those categories by toggling on “include sensitive topics in memory.” The app will notify users whenever a sensitive topic is saved.
The company also said Claude will never save some items, including government-issued ID and Social Security numbers, criminal history, immigration status or other information that would violate its acceptable use policy.
Availability
Anthropic said the memory feature is enabled by default on Free, Pro, and Max plans across web, desktop and mobile. On iOS and Android, users must update the app to the latest version to get the new memory behavior.
Original source: TechCrunch AI