
Now loading...
Beginning today, users will find that the memory capabilities in Claude have been unified across the chat and Cowork functionalities. This change means that Claude builds on existing knowledge about users, allowing for a seamless experience whether in conversation or collaborating on tasks. Users can view, edit, or delete any information that Claude retains about them, organized by topics. Notably, Claude refrains from storing sensitive information such as health details or personal beliefs by default; however, users have the option to enable memory for these topics if they wish to reduce repetitive explanations.
With the new memory features, Cowork can now leverage the same memory used in chat, resulting in enhanced efficiency and continuity. When transitioning between chat and Cowork, the context gathered over time—such as quarterly objectives and project statuses—will carry over seamlessly. For example, if Cowork is tasked with drafting a project update, it will already be aware of the intended audience and their preferences. Similarly, when brainstorming logistical details for an event in chat, Cowork can access essential elements like the attendee count and location when creating relevant documents.
Claude now actively adds topics to memory during conversations rather than compiling summaries post-discussion. If a user mentions a change, such as a project deadline being pushed to September, the updated information will be integrated into subsequent chats without any need for reminders. Users have full control over this feature, including options to pause or reset memory whenever necessary.
All stored memories are accessible through a simple interface under Topics in the Memory settings. Each entry is concise, allowing users to view, amend, or remove instances of retained information easily. For instance, correcting an outdated company name in memory will ensure accuracy across future dialogues, eliminating the need for multiple updates.
While Claude typically avoids saving sensitive information—like health or identity matters—users can choose to allow memory on certain sensitive topics if they find it beneficial. For instance, a gluten allergy could be remembered to enhance meal planning suggestions. When users opt to include sensitive topics, Claude will notify them each time relevant information is saved moving forward. Previous data will not be retroactively included, and the memory feature can be toggled off at any time.
To safeguard user privacy, Claude will not store critical identification information, such as Social Security numbers or criminal records, despite the allowance for sensitive topics in memory. Notifications will be provided if a memory update fails for any of these sensitive subjects. For additional information, users can refer to the Help Center.
Memory functionality will be activated by default for all users on Free, Pro, and Max plans across the web, desktop, and mobile platforms. It’s important to note that remembering sensitive topics is disabled initially. For users on iOS and Android, updating to the latest version of the application will ensure access to these features. Furthermore, in Team and Enterprise tiers, administrators have the authority to control memory settings for their organization, with individual users needing to activate the functionality manually.
For those interested in adjusting their memory settings, visit Settings > Memory to manage what Claude recalls.
