Anthropic integrates Claude chat and Cowork memory systems

Aug 25, 2026

On August 25, 2026, Anthropic merges the memory systems of its Claude chatbot and Claude Cowork desktop agent, allowing users to seamlessly share context across platforms. The update enables two-way memory persistence, meaning Claude dynamically retains information during active chats rather than summarizing afterward. While the feature is enabled by default for Free, Pro, and Max tiers, Anthropic excludes sensitive personal topics by default and prohibits the storage of highly private data like government IDs.

Claude memory system integration

▪For Enterprise and Teams tier organizations, Anthropic's merged memory features are turned off by default and subject to eligibility
▪On August 25, 2026, Anthropic announced the merger of the memory systems for its Claude chatbot and Claude Cowork desktop agent
▪To use the merged memory features on iOS and Android devices, users must update their Claude mobile applications to the latest version
▪The merged memory feature is enabled by default on Anthropic's Free, Pro, and Max plans across web, desktop, and mobile platforms
▪Under Anthropic's updated memory system, Claude adds topics to its memory dynamically as a user chats, rather than summarizing the conversation after it ends

Cross-product memory persistence

▪The memory integration between Claude chat and Claude Cowork is two-way, meaning actions and information from Claude Cowork are remembered in Claude chat, and vice versa
▪The memory integration between Claude chat and Claude Cowork allows the agent to access a user's existing chat history and context built up over months of prior conversations
▪Anthropic's command-line tool Claude Code is not included in this memory integration update, which focuses solely on Claude chat and Claude Cowork

Sensitive data handling controls

▪Users can view, edit, or delete specific saved memory topics globally through a new Memory section in the Claude Settings menu
▪By default, Claude does not store sensitive personal topics such as health, race, ethnicity, religion, politics, or gender identity in its memory
▪Anthropic prohibits Claude from storing highly sensitive data, including government-issued identification numbers, criminal history, and immigration status, even if sensitive memory is enabled
▪If a user enables sensitive topic storage in Claude, the AI will display a notification whenever it detects and saves a sensitive subject
▪To prevent Anthropic from using Claude conversation data for model training, users must manually toggle off the 'Help Improve our AI models' option in the privacy settings

AI product consolidation trend

▪The consolidation of memory systems between Claude chat and Claude Cowork aims to resolve a common user frustration in AI agents by eliminating the need to repeatedly rebrief the AI across different tools
▪The merger of Claude's memory systems reflects an industry trend where AI developers like Anthropic and OpenAI consolidate individual desktop applications into multi-feature 'super apps.'