Anthropic has combined the memory systems used by Claude chat and Claude Cowork. Information Claude retains from conversations can now follow users between the chatbot and the company’s task-focused Cowork tool.
David Gewirtz reports for ZDNET that the shared memory works in both directions. Context created in Cowork can also be used in later Claude chat conversations. Anthropic’s announcement does not cover Claude Code, which remains outside the scope of this update.
The change applies to existing context as well as future activity, according to Anthropic’s examples. A user who has previously discussed a manager’s preferred writing style, an event’s location, or internal business metrics could ask Cowork to prepare related material without repeating those details.
Claude will also add topics to memory during conversations. Previously, it generally created summaries after a chat ended. Users do not need to explicitly ask Claude to remember a detail for it to be considered for memory.
Controls for sensitive information
Memory is enabled by default for Free, Pro and Max users across web, desktop and mobile. Organizations using Enterprise and Teams have memory turned off by default, subject to eligibility.
Anthropic says Claude does not normally store sensitive subjects, including health, race, religion, politics and gender identity. Users can choose to allow sensitive topics in memory, for example to retain information about a food allergy. The company says it will still exclude certain data, such as sensitive identification numbers, criminal history and immigration status.
Users can pause or reset memory, turn off memory generated from chats, and review, edit or delete saved topics in Settings. The same saved topics apply across Claude chat and Cowork.
Stay up to date
AI for content creation: the latest tools, tips and trends. Every two weeks in your inbox:
