Claude's Chrome extension now shares one session across your devices

Anthropic has turned Claude in Chrome into a full Claude Cowork client: the extension's side panel now runs the same session as the desktop, web, and mobile apps. The change was announced on Threads, and it is live for Max and Team subscribers, with Pro users to follow in the coming weeks.

At a glance

• Sessions now live with the account rather than a single device, so a task started in a browser tab can be picked up later in the desktop, web, or mobile app.
• User skills and connectors are available in the Chrome sidebar, and every conversation lands in the shared history instead of staying isolated inside the extension as it was before.
• On Enterprise plans the extension ships disabled and administrators can switch it on for approved domains only, while the browser agent inherits the auto mode Anthropic uses to screen risky actions.

Why it matters The extension had been the odd surface out in the Claude line-up, useful next to a page but disconnected from everything else. Folding it into Cowork appears to make Cowork the default way of working with Claude rather than one mode among several, and it likely raises the value of skills and connectors, which now follow a user into the browser. It also widens the surface on which prompt injection can do damage.

Skills and connectors now run inside the Chrome side panel

The side panel now opens the same Claude Cowork session that runs in the desktop, web, and mobile apps. Conversations are saved to the account rather than to a single device, so work started in a tab can be resumed elsewhere, and the skills and connectors a user has configured are available in the browser.

Anthropic presents the browser as a route into systems that have no direct integration with Claude. "Many of the tools you use every day connect directly to Claude, but others don't, such as internal dashboards, legacy systems, and vendor portals," the company writes, and says Claude can work in those apps through the browser.

Auto mode extends to the browser agent, and Anthropic says the risk is not closed

Anthropic acknowledges that browser agents can be steered by instructions hidden in a page, and that Claude in Chrome carries the same prompt injection risk as any agent acting in a browser. With this release the company extends its auto mode to the browser agent, which screens actions for anything unusual instead of asking for approval at every step.

Claude still stops to ask before making a purchase or sharing personal data, and Anthropic publishes a short list of habits it recommends users keep on their own side. The company is explicit that the protections are partial, and says so in its own wording.

While these measures meaningfully reduce the risk, they cannot eliminate it.

Max and Team have it now, Enterprise ships it off by default

The updated extension is live for subscribers on the Max and Team plans, with Pro users due to get it in the coming weeks. On Enterprise plans it is disabled by default; administrators can turn it on and confine it to a list of approved domains.

Distribution stays narrow: Anthropic notes that the extension does not run on other Chromium-based browsers or on mobile. The step follows the expansion of Cowork itself, which, according to The New Stack, reached the web and mobile apps only a few weeks ago after starting as a desktop-only feature. The outlet describes that work as likely groundwork for bringing Cowork into the extension.

What's next Anthropic gives no date for the Pro rollout beyond the coming weeks, and the extension remains tied to Chrome on the desktop for now. The open question is the relationship between Cowork and ordinary Claude chat: The New Stack argues the two may eventually merge, since Cowork can do everything a basic chat does and more, which would remove a recurring source of confusion about which mode to use.