Claude's memory works everywhere, and you decide what's in it
02:00 · August 25, 2026 · Claude Blog
Summary
Claude now maintains a single memory store that operates identically in ordinary chat sessions and in Claude Cowork, the service’s task-oriented environment. Context accumulated over months of conversations—such as project deadlines, team metrics, or recurring stakeholder preferences—becomes immediately available when a user hands a task to Cowork, and any updates generated inside Cowork are reflected back in subsequent chats without additional prompting.
Memory entries are created incrementally during conversation rather than through post-hoc summarization. When a user mentions a changed deadline or revised definition of success, the system records the relevant topic in real time; the next exchange or Cowork run already incorporates the updated fact. Users can inspect every stored topic as a short, editable file under Settings > Memory, where individual entries can be revised or removed. A single correction, such as an outdated company name, propagates across all future interactions.
By default the system excludes topics classified as sensitive—health information, religious or political beliefs, race, gender identity and similar categories—from persistent storage. An optional setting allows these subjects to be retained, in which case the user receives an explicit notice each time such a topic is saved. Even with the setting enabled, certain categories remain permanently excluded, including government identification numbers, criminal history and immigration status. On Team and Enterprise plans, administrators decide whether memory is available to individual accounts; on Free, Pro and Max plans the feature is enabled by default while sensitive-topic retention remains off.
Why it matters
Provides actionable details on implementing consistent AI memory in workflows, relevant for product teams building with Claude in the Netherlands, with strong emphasis on privacy controls aligning with EU regulations.
