Claude's Memory Now Spans Chat and Cowork: What You Said in Conversation Carries into the Work, On by Default

Anthropic merged the memory systems behind Claude chat and Claude Cowork on August 25. Context built in chat previously did not carry into Cowork; now it flows both ways — Cowork can use what Claude remembered from conversations while running cloud tasks, and what Cowork gathers flows back into chat. Memory formation also changed, from summarizing after a conversation ends to maintaining a live record of topics as you chat. What Claude remembers appears as short files under Topics in memory settings, each viewable, editable and deletable. Sensitive topics — health, race, ethnicity, religion, politics, gender identity — are not saved by default, and some data such as Social Security and government ID numbers is never stored. It covers Free, Pro and Max on web, desktop and mobile, on by default for consumers; Claude Code is not included.

Two Separate Memories, Now One

Before this change, chat and Cowork were memory islands: preferences, project background and working relationships you established in chat were unknown to Cowork when it went off to run a cloud task. Now they connect in both directions, and Anthropic's examples are concrete — Cowork can draft an update already knowing who your manager is and how that person prefers updates written; a conference budget and logistics document can already include the headcount, city and speakers you discussed earlier in chat. A second change concerns timing. Claude used to wait until a conversation ended and then summarize it; now it maintains a live record of topics as you talk. The practical effect is that something you just said can be used downstream without waiting for the conversation to wrap up.

The Controls Matter More Than the Feature

What gets remembered is not a black box. Everything appears as a set of short files under Topics in memory settings, each one individually viewable, editable or deletable, with changes applying to future conversations. Anthropic's example: correct a company's outdated name in one topic and Claude uses the updated version going forward. Memory can be paused or reset at any time, through two toggles — one governing whether memory is generated from chats, the other governing sensitive topics. The privacy defaults are the more solid part of this release. By default Claude does not save sensitive topics such as health information, race, ethnicity, religious beliefs, politics or gender identity. Users who want it to can enable "include sensitive topics in memory," after which Claude displays a notice whenever it saves sensitive content — and anything discussed before the setting was enabled is not saved retroactively. A separate category is never stored at all, including Social Security numbers and government identification numbers.

The Boundaries

Coverage is stated plainly: available on Free, Pro and Max across web, desktop and mobile, on by default for consumers. Team and Enterprise customers will get rollout details separately. Claude Code is not included. "On by default" is the part of this worth pausing on. The feature saves most people effort, but it means that if you do nothing, cross-context accumulation is already happening. The two minutes worth spending: open memory settings and read the Topics list — see what Claude has so far recorded as facts about you, particularly the things you mentioned in passing and would rather it not carry around indefinitely.

via: TechCrunch, Engadget, The Next Web, 9to5Mac