Anthropic Unifies Claude Memory Across Chat and Cowork
Anthropic announces a major update to Claude's memory system that eliminates one of the most frustrating aspects of using AI agents: constantly re-explaining information the assistant already knows. The company merges the memory used by Claude's chat interface and Claude Cowork, so anything Claude learns in one area carries over to the other. Anthropic also exposes what Claude retains, letting users read, edit, or delete memories on any topic.
The change makes Claude feel like a single continuous assistant rather than two separate products under one roof. Previously, users could develop project ideas over long chat conversations but had to start over and rebrief Cowork when it came time to act. Now, for example, when drafting a conference agenda or manager update, Cowork already understands references like headcount, city, and speakers from past chats. Claude also adds topics to memory as conversations unfold rather than summarizing them at the end, so memories update and sync between chat and Cowork even mid-conversation.
Privacy protections remain central to the design. By default, Claude does not store personal or sensitive information such as health data, race, ethnicity, religious beliefs, politics, or gender identity, though users can toggle on the inclusion of sensitive topics if they prefer. The app notifies users whenever a sensitive topic is saved, and some data—like government IDs, Social Security numbers, criminal history, and immigration status—is never stored. The feature is enabled by default on Free, Pro, and Max plans across web, desktop, and mobile, with iOS and Android users needing to update their apps to the latest version.