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 shifted its default settings so that memories about users are shared automatically between Claude, its chat assistant, and Cowork, its task-oriented AI. This change impacts user privacy and convenience, with users needing to opt out if they prefer separation. For more details, see the original analysis on privacy implications of shared AI memories.
Anthropic has enabled shared memory by default between its two AI products, Claude and Cowork, meaning that information about a user’s preferences, ongoing projects, and personal details will now automatically transfer across both tools unless the user actively opts out. This shift in default settings significantly alters the privacy landscape for users who rely on these AI services for personal or professional tasks.
The company announced that memory sharing is now the default for users who operate both Claude and Cowork with memory enabled. Previously, users had to manually enable sharing, giving them control over what information was transferred. Under the new setup, memories collected in Claude—such as user preferences or project details—are accessible to Cowork, an agent designed to perform multi-step work tasks like research, drafting, and coordination.
Anthropic clarified that users can disable memory sharing in Claude’s settings, but the default change means most users will experience automatic transfer unless they take action. The announcement did not specify whether existing memories are migrated automatically or if users can selectively prune memories before sharing. It also remains unclear how the shared data is used beyond facilitating task execution, or whether it is incorporated into model training.
Implications for User Privacy and Convenience
This change raises important privacy considerations because default sharing means personal and professional information stored in Claude will now be accessible to Cowork without explicit user consent. For users who value privacy, this could lead to unintended exposure of sensitive data. Conversely, supporters argue that this setup reduces repetitive input and streamlines workflows, especially for those who rely on both products for work and personal tasks.
The move also reflects a broader industry trend where AI providers are increasingly integrating memory features that persist across sessions, prompting ongoing debates about data retention, user control, and transparency. As competitors like OpenAI and Google expand their memory capabilities, Anthropic’s default sharing policy highlights the balancing act between convenience and privacy in AI development.
As an affiliate, we earn on qualifying purchases.
Background on Anthropic’s Memory Features
Anthropic introduced memory for Claude as a way to let the assistant retain information across conversations, such as user preferences, recurring projects, or technical environments. The feature was designed to be transparent, allowing users to review and delete stored memories. Cowork, on the other hand, is part of Anthropic’s push into agentic AI—assistants that perform multi-step tasks rather than just answering questions.
Previously, users had to enable memory sharing intentionally, giving them control over what context was transferred between the chat and task-oriented functions. The recent change shifts this control, making shared memory the default, aligning with Anthropic’s goal of creating more integrated, capable AI tools that work seamlessly in tandem.
“Users can opt out of memory sharing between Claude and Cowork in settings.”
— Anthropic spokesperson
Details Still Unclear on Memory Management
Several key questions remain unanswered: It is not confirmed which specific types of memories are shared automatically, whether existing memories transfer without user intervention, or if users can selectively control what is shared. The exact handling of sensitive or incidental data, and whether shared memories are used for model training, also remains unspecified. Additionally, rollout timing varies, so some users may not see the new default or notification immediately.
Expected Clarifications and User Controls
Anthropic is likely to provide further details on how memory sharing can be managed, including options for selective sharing and data deletion. Users should check their settings for memory options and review stored memories to ensure privacy preferences are met. The company may also clarify how shared data is used in training and whether deletion in one product affects the other. Watch for updates as the rollout continues and competitors respond to evolving privacy expectations in AI.
Key Questions
Yes, users can disable memory sharing in Claude’s settings, although it is now off by default. This allows you to keep your data separate across the two products.
What types of memories are shared automatically?
Anthropic has not specified exactly which memories are shared, but it generally includes preferences, ongoing projects, and personal details stored in Claude that could assist Cowork in performing tasks.
Will shared memories be used for model training?
Anthropic has not confirmed whether shared memories are used for training or if they are only retained for immediate task performance. Further details are expected.
How can I review or delete my memories?
You can review and delete stored memories through Claude’s memory management settings, which are accessible in the app’s interface.
When will all users see this change?
The rollout is staggered, so some users may see the default sharing setting and notification in the coming weeks. Full deployment is ongoing.
Source: ThorstenMeyerAI.com
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
