Anthropic unifies Claude chat and Cowork memory systems to reduce rebriefing friction

Anthropic announced a unification of the memory systems used by its Claude chat and Claude Cowork products, enabling the AI to retain context across both modes. Previously, users had to rebrief Cowork on details already discussed in chat, which disrupted workflows when moving from research to action. The merged memory system now ensures that what Claude learns in one context is available in the other, making the assistant feel like a continuous experience rather than two separate products.

The company also introduced user controls for memory, allowing users to read, edit, or delete information Claude has retained. By default, the system excludes sensitive topics such as health data, race, ethnicity, religious beliefs, politics, and gender identity, though users can opt in to include these categories. Anthropic states it will never store government-issued IDs, Social Security numbers, criminal history, immigration status, or other data that violates its acceptable use policy.

Anthropic noted that memory updates now occur as conversations progress, rather than only when a chat ends, enabling faster sharing of new context between chat and Cowork. The feature is enabled by default for Free, Pro, and Max plans across web, desktop, and mobile, with mobile users required to update to the latest app version to access the change.