Anthropic is making Claude’s memory more consistent across its growing range of AI tools, with a new update that connects what the assistant remembers in regular Claude chats with its Cowork environment.
The change means Claude now uses a shared memory across chat and Claude Cowork. Information saved while talking to Claude can therefore be available when handing a task to Cowork, while relevant details learned through Cowork can carry back into future conversations. The feature is designed to reduce the need for users to repeatedly explain projects, preferences and other ongoing context when moving between different parts of Claude.
That continuity could become increasingly important as AI assistants evolve beyond standalone chatbots. Tools such as Cowork are intended to handle more involved tasks rather than simply answer questions, making persistent context potentially more useful. An assistant that already understands the background of a project, for example, can theoretically begin working with less setup each time.
It also makes memory controls more consequential. The more places an AI assistant can carry information between, the more important it becomes for users to understand exactly what has been retained and where that information can appear.
Anthropic says users remain in control of Claude’s stored memory. The company is also treating sensitive information differently from ordinary preferences and project details. Topics it considers sensitive, including health information and beliefs, are not saved to memory by default. Users who specifically want Claude to remember that type of information can enable the option through Claude’s Memory settings.
The approach attempts to balance convenience with privacy, although the effectiveness of that balance will depend heavily on how understandable and accessible those controls are in everyday use. Persistent memory can make an AI assistant considerably less repetitive, but it also introduces another layer of data management for people who may not always remember what they previously allowed an assistant to retain.
Claude memory is available across Anthropic’s Free, Pro and Max plans and works on web, desktop and mobile. Anthropic says memory is enabled by default on those plans, while the separate option for saving sensitive topics remains disabled unless users choose otherwise. Mobile users on iOS and Android will need the latest version of the Claude app to receive the newest memory changes.
The update also brings Anthropic closer to the broader idea of an AI assistant that maintains useful context regardless of how a user interacts with it. As AI companies expand from conversational interfaces into agents capable of performing longer and more autonomous tasks, continuity between those experiences is likely to become a bigger competitive feature.
For Claude users, the immediate benefit is straightforward: fewer repeated explanations when switching between chat and Cowork. The more important question over time will be whether shared memory can remain transparent and manageable as Claude is trusted with increasingly complex work.
