Anthropic has changed how Claude remembers you. The company announced an update that unifies memory across Claude Cowork and regular Claude chat, so what you tell the AI in one place carries over to the other.
According to TechCrunch, the point is to spare users from repeatedly briefing the AI on projects, preferences, and other context — a chore anyone who has restarted a chat session will recognize. 9to5Mac frames the same change as a unification of a memory feature that previously lived separately in Cowork and chat.
The rollout is broad. CNET reports that memory is enabled by default across the free, Pro, and Max plans, and works on web, desktop, and mobile. CNET also notes that Anthropic is updating the Claude mobile app as part of the change.
There's a privacy dial attached. CNET reports that sensitive topics are off by default and must be manually enabled — meaning Claude won't quietly build a running file on the more personal things you discuss unless you tell it to. SiliconANGLE describes the update as improving both customization and the handling of sensitive topics, suggesting the two goals are being shipped together rather than traded off.
Why this matters: memory is quietly becoming the thing that separates a chatbot you use from an assistant you rely on. An AI that forgets everything between sessions is a search box with better manners; one that remembers your projects and working style starts to behave like a colleague. But persistent memory is also a persistent record, which is why the default-off switch for sensitive topics is the detail worth watching — it's Anthropic's attempt to make Claude more useful without making it feel like it's keeping notes on you.