Anthropic unifies Claude memory to work across Cowork and your chat

Anthropic has confirmed that Claude’s chat memory now works directly with Cowork. That also means conversations can now carry useful context into cloud-based Cowork tasks. Users also get clearer control over what Claude remembers.

Claude’s memory now follows you

Cowork can use context built through regular Claude conversations. Work completed through Cowork can also carry relevant context back into chat.

That could remove a frustrating part of AI workflows. Users no longer need to repeatedly explain projects, deadlines, preferences, or team details.

Anthropic says Claude now updates memory during conversations. Previously, memory was summarized after conversations ended. Memories appear as individual files under Topics inside Memory settings. Users can open, edit, delete, pause, or reset them.

Sensitive memory remains an important choice

Anthropic says sensitive subjects remain excluded from memory by default. These include health, race, ethnicity, religion, politics, and gender identity.

Users can enable sensitive-topic memory through settings. Claude then shows a notice whenever it saves sensitive information going forward. The company also says some information stays excluded regardless of that setting. This includes government identification numbers, criminal history, and immigration status.

Memory is enabled by default for Free, Pro, and Max users across web, desktop, and mobile. Team and Enterprise admins control availability. At a time when AI assistants are becoming more persistent, memory control is becoming increasingly important. Claude’s approach gives users visibility instead of hiding everything behind one switch.