Claude update
Review Claude memory topics before they shape work
Memory can save time, but only when the remembered context is visible, editable and still true.
What changed
Anthropic's August 25, 2026 release notes say memory now works across Claude chat and Claude Cowork in the cloud. Remembered items are listed as Topics in Settings, Memory, where users can edit or delete them.
Anthropic also says sensitive topics such as health or beliefs stay out of memory unless the user turns on the setting to include sensitive topics. Memory is on by default for Free, Pro and Max plans, and off by default for Team and Enterprise organizations.
Try this
- Open Claude Settings, then Memory.
- Review Topics before starting recurring or client-facing work.
- Edit anything that is too broad, outdated or no longer true.
- Delete temporary project details that should live in a project brief instead.
Follow-up prompt
"Before using memory for this task, tell me what assumptions you are using about my preferences, project context or output format."
Make it easier
Use memory for stable working preferences, such as tone and format. Keep temporary facts in the project itself. That separation makes wrong context easier to spot.
Check before relying on it
Memory defaults and admin controls differ by plan and organization type. Teams should decide what belongs in memory before using it for repeated work.
Source
- Anthropic Help: Claude release notes
- Date checked: 2026-09-02