Anthropic announced that Claude now uses a single shared memory layer across its chat interface and Claude Cowork, letting users control what personal context is retained. When you delegate work to Cowork, the agent can draw on knowledge already gathered from prior conversations, such as ongoing projects, a manager's stated preferences, or client details from earlier quarters. The rollout also adds user-facing memory controls: a dedicated Memory tab, topic categories for stored memories, and a Settings toggle for whether sensitive topics should be remembered. Anthropic frames this as user-directed continuity rather than silent background profiling. The packet does not specify availability by plan, region, or rollout timing, so teams should verify what appears in their own accounts before changing workflows.
Anthropic says Claude now has one memory across chat and Claude Cowork, and you decide what is in it. Hand Cowork a task and it starts from what Claude already knows from your chats: the project you talked through, your manager's preferences, or the client from last quarter.
Key takeaway
Claude Cowork can now start delegated tasks with context pulled from your chat history, under user-controlled unified memory.
What happened
Anthropic says Claude now has one memory across chat and Claude Cowork, and you decide what is in it. The change is framed as a single continuity layer rather than separate silos for conversational chat and task-oriented Cowork work.
Hand Cowork a task and it starts from what Claude already knows from your chats: the project you talked through, your manager's preferences, or the client from last quarter. The update also adds a dedicated Memory tab, separate memory topic categories, and a Settings toggle to keep memories about sensitive topics.
Evidence
Anthropic unified Claude memory across chat and Claude Cowork with user control over stored context.
X · attributed
Anthropic says Claude now has one memory across chat and Claude Cowork, and you decide what is in it.
Cowork tasks can start from prior chat context such as projects, manager preferences, or past clients.
X · attributed
Hand Cowork a task and it starts from what Claude already knows from your chats: the project you talked through, your manager's preferences, or the client from last quarter.
Claude added a Settings toggle for sensitive-topic memories, categorized memory topics, and a dedicated Memory tab.
X · attributed
Also, a new toggle to keep memories about sensitive topics has been added to the Settings menu.
Why it matters
Shared memory across chat and Cowork can cut repeated context handoffs for agentic work, but it makes explicit user choices about stored and sensitive information more consequential.
Limits and uncertainties
The packet does not state rollout timing, plan eligibility, or regional availability for unified memory.
The packet does not describe how memory is stored, synced, deleted, or enforced across devices and sessions.
Practical implications
Review Claude Memory settings, including the sensitive-topic toggle and categorized topics, before delegating Cowork tasks that rely on prior chat context.
Treat chat conversations as potential inputs to Cowork execution and avoid storing context you would not want reused in later delegated work.
What to watch
Whether unified memory appears in your Claude account and whether Cowork tasks actually reuse prior chat context as described.
How Anthropic documents retention, deletion, and controls for sensitive-topic memories after the Settings toggle ships.
Original reporting: Claude now has one memory across chat and Claude Cowork, and you decide what's in it.
Hand Cowork a task and it starts from what Claude already knows from your chats: the project
