TL;DR
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
Anthropic has enabled shared memory between Claude and Cowork by default, meaning information the assistant remembers about a user will carry across both products unless the user opts out. The change raises convenience and privacy trade-offs for people who use Anthropic’s tools.
Anthropic will now share memories about users across Claude and Cowork by default, meaning details the assistant has learned about a person will follow them between the company’s chat assistant and its agentic work tool unless they actively opt out. The change shifts the privacy default for Anthropic’s memory feature, putting the burden on users who want their context kept separate to change a setting. It affects anyone who uses both products with memory enabled.
Under the new arrangement, memories that Claude accumulates — such as a user’s stated preferences, ongoing projects, or personal details shared in conversation — will be available to Cowork, Anthropic’s agent designed to carry out multi-step work tasks. Previously, memory was scoped more narrowly, and users had to choose to share context between products. Now sharing is the default, and Anthropic is notifying users that they can disable it.
The opt-out mechanism lives in Claude’s settings, according to Anthropic. Users who do not want their memories to flow between the chat assistant and Cowork can turn off sharing there. Anthropic has framed shared memory as a convenience feature: a user who tells Claude about their workflow, tools, or priorities should not have to repeat themselves when they hand a task to Cowork.
It is not yet clear from the announcement exactly which categories of memory are shared first, whether existing memories migrate automatically into Cowork, or whether shared memory can be selectively pruned per product rather than only toggled on or off globally.
Privacy Defaults Shift to Sharing
The decision matters because default settings shape behavior at scale. Most users never change defaults, so moving from opt-in to opt-out effectively makes cross-product memory sharing the norm for the bulk of Anthropic’s user base. For supporters of the change, that means less repetition and more capable assistance; for privacy-conscious users, it means personal details collected in one context now follow them into another without an affirmative choice.
The change also lands amid a broader industry debate about AI memory. Rivals including OpenAI’s ChatGPT and Google’s Gemini have both expanded memory features that persist information across conversations, and each has faced questions about how remembered data is used, retained, and exposed to other people who share a device or account. Anthropic’s move tightens the integration between a conversational assistant and an agent that takes actions, which raises the stakes: a memory that informs a chat reply is one thing, a memory that informs an autonomous task another.
privacy-focused AI memory management tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
How Claude’s Memory Evolved
Anthropic introduced memory for Claude to let the assistant retain information across conversations, such as a user’s writing style, technical environment, or recurring projects, rather than starting each chat from scratch. The feature is designed to be inspectable: users can review what Claude remembers and delete individual memories.
Cowork is Anthropic’s entry in the fast-growing category of agentic AI — assistants that don’t just answer questions but carry out multi-step work such as researching, drafting, and coordinating tasks. Shared memory is a natural technical step for such products, since an agent performs better when it knows the user’s context. The new default links these two pieces of Anthropic’s product line more tightly together, in line with the company’s push from chatbot toward AI that does work on a user’s behalf.
“Users can opt out of memory sharing between Claude and Cowork in settings.”
— Anthropic
AI assistant privacy settings device
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
What Anthropic Hasn’t Specified
Several details remain unclear. Anthropic has not fully spelled out which memory types transfer to Cowork and whether sensitive or incidental details are filtered. It is not confirmed whether users can share memories selectively — for example, keeping professional context in Cowork while excluding personal details — or whether the control is a single on/off toggle.
Rollout timing also appears staggered, so some users may not see the setting or the notification immediately. Whether shared memories are used for model training, and how deletion in one product affects the other, are additional open questions Anthropic has not addressed in detail in the announcement.
secure data storage for AI applications
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Checking Your Settings This Week
Users of both Claude and Cowork who want their memories kept separate should check Claude’s settings for the memory-sharing option and review the memories list to delete anything they do not want carried across products. Anthropic is expected to surface notifications about the change as it reaches accounts.
Watch for further clarification from Anthropic on selective sharing, retention, and training use — and for how rivals respond as memory becomes a competitive battleground among AI assistants. Users on shared or work-managed accounts should also verify whether administrators control the setting on their behalf.
AI memory privacy protection software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What exactly is being shared between Claude and Cowork?
Memories Claude has saved about you — preferences, project details, and other context from conversations — will be available to Cowork so the agent can act with that background. Anthropic says you can opt out in settings.
How do I stop Claude and Cowork from sharing memories?
Go to Claude’s settings and disable memory sharing. You can also review and delete individual memories from the memory list to remove details you don’t want retained.
Does opting out delete my memories?
No. Opting out stops sharing between the products; memories may still exist within Claude itself. Delete individual memories if you want them removed entirely.
Why did Anthropic make sharing the default?
Shared context lets Cowork perform tasks without users repeating information they’ve already told Claude. Anthropic frames it as a convenience improvement, though it shifts the privacy default from opt-in to opt-out.
Are shared memories used to train models?
Anthropic’s announcement does not fully address this. Users concerned about retention and training use should check the company’s privacy documentation, as details remain unclear.
Source: Anthropic
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
