Anthropic, the company behind the Claude chatbot, has rolled out new memory management features that give users more control over what information the AI remembers about them. Since the summer of 2025, Claude has been able to recall past interactions, a feature known as "memory," which allows the chatbot to provide more personalized responses and avoid repeated explanations. This feature is now enabled by default for all users, including those with free accounts, and a recent update lets the chatbot carry its memory across different conversations and sessions in its collaborative workspace, called Claude Cowork. Users can now edit or delete specific topics that Claude remembers about them through the web version of the app at claude.ai or via the mobile app. On the web, this is done through the "Memory" section under Settings, while on mobile, users can access the "Memory files" section under App settings. Anthropic has also introduced an option called "Include sensitive topics in memory," which allows Claude to record information related to health, race, ethnicity, religious beliefs, politics, or gender identity. However, users will be alerted each time Claude generates a memory about a sensitive subject. This notice may not appear on older versions or the mobile app, and in those cases, sensitive topics will not be saved. To stop Claude from automatically creating memories, users can disable the feature by turning off "Generate memory from chats." They can choose to pause memory or reset it entirely. However, resetting memory is irreversible—once turned back on, Claude will start fresh without any prior information. To keep conversations organized, Claude’s memory is separated by projects, which helps prevent the chatbot from getting overwhelmed with unrelated information. Users can add comments, documents, or GitHub links to a project to guide Claude’s memory, and at the end of each day, the chatbot will summarize the work done, which can be edited through the same interface. For information that needs to be remembered across all tools, users can add it to the "Profile" section of the Memory menu. If users want to temporarily stop Claude from recording information, they can use "Incognito mode" by clicking the ghost icon. In this mode, chats are not saved, added to memory, or used for training. Earlier this year, Anthropic introduced a feature that lets users import conversations from other chatbots into Claude, further expanding its ability to learn from and adapt to different platforms.