Anthropic Unifies Memory Between Claude Cowork and Claude Chat

Anthropic has merged the memory systems behind Claude Cowork and Claude chat, allowing the AI to draw on context from either surface regardless of where a conversation started.

What changed

Previously, Claude Cowork — Anthropic's cloud-based task execution tool — and Claude's standard chat interface operated with separate memory. Under today's update, both draw from the same underlying memory store.

"Cowork now has memory, and it's the same one you use in chat, leading to less re-explaining and more picking up where you left off," Anthropic said in its announcement. "When Cowork runs a task in the cloud, what Claude remembers from your chats is there, and vice versa."

In practice, this means a user who discussed a project, a manager's preferences, or a client relationship in chat can hand that context to Cowork without re-explaining it, and information Cowork gathers while executing a task becomes available in future chat sessions.

Sensitive topics remain restricted by default

Anthropic says memory does not automatically retain sensitive categories such as health information or personal beliefs. According to the company, "Claude does not store subjects considered sensitive, like health or beliefs, to memory by default, but if those are the topics you're tired of re-explaining, you can turn them on in Memory settings." Users can manually enable storage of these categories through Memory settings if they choose.

Availability

Anthropic says memory is on by default across Free, Pro, and Max plans, spanning web, desktop, and mobile. The company is directing iOS and Android users to update to the latest app version to receive the change. Pricing for Claude plans was not altered as part of this update, and no new model or version was introduced — this is a feature change to how existing Claude models handle stored context across products.

What this means

This is an infrastructure change to Claude's product surface, not a new model release. Anthropic is treating memory as a cross-product utility rather than a feature siloed to individual tools — a design choice that mirrors how competitors like OpenAI have expanded ChatGPT memory across its own web, mobile, and enterprise surfaces.

The practical effect is reduced friction for users who move between quick chat sessions and longer-running Cowork tasks, since context no longer needs to be manually re-supplied. The default exclusion of sensitive topics — health and beliefs specifically named — signals Anthropic is trying to get ahead of privacy concerns that have dogged memory features industry-wide, though the opt-in toggle means the actual privacy posture depends entirely on user configuration rather than a hard technical boundary.

No benchmark data, model version, or pricing changes accompanied this announcement, indicating this is a product-layer update to Claude's existing models rather than a change to model capability or cost structure.