Claude can now recall past chats, but only when you ask

Anthropic has introduced a memory feature for its Claude chatbot that lets users pull in context from earlier conversations on demand. In a demo video, a user asks what they were working on before a vacation, and Claude searches previous chats, summarizes the thread, and offers to continue the project. The tool works across web, desktop, and mobile, and can keep projects and workspaces separate.  It begins rolling out today to Max, Team, and Enterprise plans, with broader availability promised. To enable it, users can open Profile → Settings and switch on “Search and reference chats.” Unlike persistent memory in some rivals, Claude will not remember by default or build a user profile; it retrieves and references past chats only when asked, according to Anthropic spokesperson Ryan Donegan.  The launch lands amid rapid one-upmanship between Anthropic and OpenAI, with both pushing new features such as voice, larger context windows, and new tiers while raising significant funding. OpenAI released GPT-5 last week, and Anthropic is seeking a round that could value the company at up to $170 billion.  Memory is seen as a way to make chatbots stickier by saving users from re-explaining work and maintaining continuity across long projects. The move also comes as chatbot memory draws attention online: some users praise the convenience of continuity, while others criticize deeper recall and the ways people are using AI for sensitive situations, including reports of mental health struggles linked to heavy chatbot use.  For now, Anthropic is threading a middle path, offering recall when requested, not by default, positioning Claude as a tool that helps you pick up where you left off without quietly collecting details in the background.