AnswersPublished22 sources
Claude and Cowork Now Share Memory: What Anthropic Changed
As of August 25, 2026, Claude chat and Cowork use the same two way memory: context learned in one can carry into the other, including months of conversation history and work preferences. Claude can now add and update memory while a conversation is happening instead of waiting until the chat ends.
Published byEdited with GPT-5.6 LunaImages generated with GPT Image 1.5
Research answer
Key takeaways
- As of August 25, 2026, Claude chat and Cowork use the same two way memory: context learned in one can carry into the other, including months of conversation history and work preferences.
- Claude can now add and update memory while a conversation is happening instead of waiting until the chat ends.
- Sensitive topics such as health, race, ethnicity, gender identity, religion, and political beliefs are excluded by default, though users can opt in to retaining them.
AI Prompt
openai.comCreate a landscape editorial hero image for this Studio Global article: What did Anthropic’s update to Claude’s memory system change about how memories work across Claude chat and the desktop AI agent Cowork—incl. Article summary: Anthropic merged Claude chat’s and Cowork’s memories into one two-way system: context learned in either product can be used in the other, reducing the need to re-explain ongoing work. The update is enabled by default, su. Topic tags: general, general web. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, charts with fake numbers, clic
Anthropic has merged the memory systems behind Claude chat and Claude Cowork into one two-way system. Information Claude retains from chat can now be available when you hand a task to Cowork, while useful context learned during Cowork can flow back into chat. The practical change is continuity: users should need to spend less time re-explaining ongoing work, preferences, and project context. 18
20
22
What changed in Claude’s memory
Previously, Claude chat and Cowork maintained separate memory experiences. The update makes them function more like a single assistant across supported workflows. Anthropic says the context built over months of conversations can follow a user into Cowork, including when Cowork runs a task in the cloud. 22
26
The flow works in both directions:
- Information remembered from Claude chat can inform a Cowork task.
- Context learned through Cowork can become available in later Claude chats.
- Users can move from discussion to execution without repeating every standing instruction or project detail.
For example, a user might explain in chat how their team defines its metrics, then ask Cowork to use that context while preparing a quarterly business review. Other durable details—such as a changed deadline or a preferred writing style—can similarly help Claude produce more relevant follow-up work when they are retained in memory. 25
This is persistent memory, not simply a longer context window. A context window holds information supplied to a particular interaction; memory is the separately retained information Claude can use across interactions. Anthropic’s platform documentation describes its API agent memory as a distinct, workspace-scoped memory store, so that developer mechanism should not be confused with the consumer chat-and-Cowork update. 17
Memory now updates during the conversation
Claude can also update memory in real time as a conversation progresses. It no longer has to wait for a conversation to finish and summarize it before adding relevant information. 19
23
24
That means a change made midway through a discussion—such as a project deadline moving to September—can be reflected in later interactions without the user explicitly saying “remember this.” The feature is designed to capture useful continuity automatically, rather than making users manage every memory with a manual instruction. 18
25
Automatic updates make memory more convenient, but they also make review important. A detail that was accurate in one conversation may later become outdated, incomplete, or too specific to apply broadly.
New controls for reviewing and deleting memories
Anthropic is making retained information easier to inspect. Users can view what Claude remembers organized by topic, then edit or delete individual memories. Memory can also be paused or reset through Settings > Memory. 19
21
These controls matter because shared memory expands the reach of retained information. A detail saved during a casual chat may later influence a Cowork task, while something learned during an agent task may affect a future conversation. Reviewing the memory list is therefore the clearest way to check what Claude may carry between the two products.
Sensitive topics are off by default
Claude does not retain sensitive subjects by default. The announced examples include health, race, ethnicity, gender identity, religion, and political beliefs. Users can choose to enable memory for those topics in Memory settings if they prefer not to repeat them. 20
21
24
The default protects against automatically creating a durable profile from highly personal information, but it is not a reason to treat Cowork like a risk-free environment. Anthropic’s Cowork guidance warns that computer use can expose information visible on the screen or in applications the user has authorized, including personal data and sensitive documents. It recommends avoiding sensitive local files and being deliberate about which signed-in websites Cowork can access. 26
37
The supplied announcement coverage does not establish a complete list of information categories that are permanently barred from memory. It is therefore more accurate to distinguish between topics excluded by default and categories that Anthropic may permanently prohibit, rather than assume the latter list.
Where the shared memory works—and where it does not
The available documentation confirms shared memory between Claude chat and Cowork when Cowork runs tasks in the cloud. Cowork sessions that run locally on the user’s computer do not use the shared chat memory. 27
Contemporary coverage says the feature is enabled by default for consumer users on Free, Pro, and Max plans across Claude’s web, desktop, and mobile experiences, subject to memory settings and exclusions. 18
20
The supplied evidence does not reliably establish identical availability for every Team or Enterprise account, every region, or every Cowork configuration. Those details should be checked in the account’s current settings rather than inferred from the consumer rollout.
What this means for Claude Code
Claude Code should not automatically be assumed to participate in the shared Claude-chat and Cowork memory system. The announcement and support material supplied here describe the integration as a chat-and-Cowork feature; they do not document Claude Code as a connected reader or writer of that same memory.
Developers do have a separate API agent-memory mechanism: Anthropic’s platform documentation describes a workspace-scoped memory store for managed agents. That is a different system with different implementation and access rules. 17
The practical takeaway
Anthropic’s update changes Claude memory from a chat-specific convenience into shared continuity across chat and cloud-based Cowork tasks. The biggest benefits are less repeated briefing and better reuse of work preferences, project context, and changing details. The biggest new responsibility is auditing what has been retained—especially because memory now updates during conversations and can influence work in another Claude product.
For a safer setup, review Settings > Memory, remove stale items, leave sensitive-topic memory off unless it is genuinely useful, and avoid giving Cowork access to sensitive files or high-consequence accounts. 19
26
37
Studio Global AI
Continue your research
This page includes a source-backed answer you can continue inside Studio Global.
People also ask
What is the short answer to "Claude and Cowork Now Share Memory: What Anthropic Changed"?
As of August 25, 2026, Claude chat and Cowork use the same two way memory: context learned in one can carry into the other, including months of conversation history and work preferences.
What are the key points to validate first?
As of August 25, 2026, Claude chat and Cowork use the same two way memory: context learned in one can carry into the other, including months of conversation history and work preferences. Claude can now add and update memory while a conversation is happening instead of waiting until the chat ends.
What should I do next in practice?
Sensitive topics such as health, race, ethnicity, gender identity, religion, and political beliefs are excluded by default, though users can opt in to retaining them.
Sources
thenextweb.comAnthropic merges Claude chat and Cowork memory, on by default
9to5mac.comAnthropic update unifies memory feature across Claude Cowork and chat - 9to5Mac
zdnet.comAnthropic's Claude and Cowork will share memories about you now - unless you opt out
support.claude.comLet Claude use your computer in Cowork