A significant update to Claude’s memory system is making the AI more user-friendly by allowing it to retain information across chat and Cowork sessions. This means that users no longer have to re-explain their ideas from scratch each time they switch between tasks.

For example, if you're drafting an agenda for a conference or an update for your manager, Claude will now understand the context of past conversations, such as headcount numbers and speaker details, eliminating the need to repeat information. This feature enhances continuity and efficiency in workflow management.

The new memory system also means that Claude will add topics to its memory as you chat, rather than just summarizing at the end of a conversation. This makes it easier to move between different tasks without losing context or starting from scratch.

However, the update comes with some caveats regarding privacy and security. Anthropic has stated that by default, personal information like health data, race, ethnicity, religion, politics, gender identity, and more will not be stored in Claude’s memory unless explicitly enabled by the user. The app also notifies users whenever sensitive topics are saved.

It's worth noting that certain highly sensitive information, such as government-issued ID numbers or criminal history, is never saved to protect user privacy and comply with acceptable use policies.