Anthropic has quietly rolled out a change that allows its Claude AI assistant and its Cowork productivity tool to share memory data about users across both products. The feature is enabled by default, meaning anyone using both tools will have their stored preferences and contextual information passed between them unless they take steps to disable it.
What the Memory Sharing Actually Does
Memory sharing allows Claude to carry context from one product into another. If Claude has learned your writing preferences or workflow habits inside Cowork, that information can now surface during a standard Claude conversation, and vice versa. The intent is to create a more consistent, personalized experience across Anthropic's growing suite of tools. In practice, it means the two products no longer operate as independent silos when it comes to what they know about you.
Key Facts
- Memory sharing between Claude and Cowork is on by default for users of both products.
- Shared data includes stored preferences, behavioral context, and personalization signals.
- Users can opt out through their account settings, though the path is not prominently surfaced.
- The change affects all users who have memory features enabled on either platform.
- Anthropic has not announced a firm timeline for expanding cross-product memory further.
The rollout follows Anthropic's recent pattern of expanding integrations between its products without making the defaults especially visible. For users who have been following Claude Cowork Now Runs as a Chrome Sidebar Extension, this is another step toward tighter product cohesion. Whether that cohesion comes at a privacy cost depends on how comfortable a given user is with AI tools pooling their behavioral data.
Users who want to maintain a clear boundary between their Claude conversations and their Cowork sessions will need to find and toggle the relevant setting themselves.ZDNET
How to Opt Out and Why It Matters
Opting out requires navigating to account settings within the Claude platform. The option is available, but it is not shown as a prompt or notification during regular use, which means many users may not realize the change has taken effect. Privacy advocates have pointed out that default-on data sharing arrangements place the burden on users to protect their own information rather than on the company to seek affirmative consent. This concern is not new to Anthropic's products. Earlier reporting on Claude Shared Chats Indexed by Google: What You Need to Know raised similar questions about how user-generated content and data surfaces in unexpected places.
The broader context here matters. Anthropic is building out a product ecosystem that includes multiple interfaces and use cases. Understanding the differences between those products, and which data flows between them, is increasingly important. Our guide on How to Pick Between Claude Models, Code, and Cowork outlines the distinct purposes each tool serves, which in turn helps clarify why a memory bridge between them has real implications for how you use each one.
A Familiar Pattern in AI Product Design
The default-on approach to cross-product data sharing is common across the AI industry, not unique to Anthropic. But it carries particular weight when the products involved are designed to learn about users over time. Memory features are meant to make AI assistants more useful, and that value proposition is genuine. The tradeoff is that the more an AI remembers and shares, the more it requires trust in the company managing that data. Anthropic has generally been forthcoming about its data practices, but decisions like this one test whether that reputation holds as the product line expands. Users should review their settings now if they have any preference about keeping their Cowork and Claude experiences separate.
