Anthropic Unifies Claude’s Memory for 3M+ Users, Sensitive Data Opt-In

- Unified memory syncs user details across Claude chat and Cowork for streamlined experiences - Sensitive topics excluded by default, with explicit user opt-in required for personal information On August 26, 2026, CoinDesk reported that Anthropic has unified the memory system across Claude chat and Cowork. This change benefits more than 3 million users by allowing seamless sharing of information between casual chats and work tasks, reducing repetitive input. The update introduces automatic recall of user-provided details across both platforms and adds controls for editing, deleting, or opting out of retained topics. Memory now operates by default for all Free, Pro, and Max subscribers where available, while Cowork remains accessible only to paid accounts. Users can review, modify, or remove remembered topics, and they also have the option to pause memory or activate incognito mode. Anthropic clarified that sensitive topics, including health, politics, or gender identity, are kept out of shared memory unless users explicitly opt in through settings. Meanwhile, social security numbers, government IDs, criminal history, or any content violating the Acceptable Use Policy are never stored. Users can no longer keep chat and Cowork memories separate unless they use different accounts, although the Claude Code developer tool maintains independent memory. For enterprise subscribers, administrators control memory feature availability, and on mobile, the latest app version is required for the update. Anthropic’s policy revamp aims to boost productivity while strengthening user privacy, and it ensures comprehensive oversight of what the AI assistant remembers.