AI 服務暫時不可用，以下為來源正文，待恢復後補全翻譯。
Anthropic announced a major update to its Chrome extension on Wednesday that turns Claude in Chrome from a useful but somewhat disconnected experience into a full Claude Cowork client in the browser with sessions that persist across Anthropic’s various apps The updated extension is now available to all Max and Team plan subscribers, with plans to bring it to Pro users in the coming weeks. Until now, Claude in Chrome was more or less its own thing and almost completely separate from the other Claude apps, making it useful in the browser context but not as a companion to the other Claude surfaces. Indeed, that’s one of the main complaints reviewers have about the extension in the Chrome Web Store. Since the extension is now deeply connected to the overall Claude ecosystem, all conversations are now saved in the user’s history, so it’s easy to start a conversation on the desktop or mobile app and pick it up in the browser, for example. What’s just as important, though, is that all of a user’s skills and connectors are now also available in the Claude sidebar on Chrome. Credit: Anthropic. The main advantage of the browser extension is, of course, that it can use the browser and see what the user is looking at. “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,” Anthropic explains. “With Claude in Chrome, Claude can work in these apps through the browser.” Claude Cowork as the default Anthropic brought Cowork to the web and mobile only a few weeks ago. That work was likely necessary to also enable Cowork in the Chrome extension. Before, Cowork only existed in the desktop app, after all. But now, it’s almost becoming the de facto standard for using Claude. Maybe Anthropic will consolidate Cowork and its basic Claude Chat over time. That would also clear up some of the confusion many users likely have about when to use regular chat or Cowork in Claude, because at the end of the day, Cowork can do everything a basic chat can — and much more. Credit: Anthropic. Safety through auto mode As Anthropic notes, “Claude in Chrome carries the same risks as any AI agent that acts in a browser, chiefly prompt injection.” If a malicious actor hides instructions for the agent on a website, that could be a problem, but with this update, Anthropic is essentially extending its Claude auto mode to the browser-based agent as well, with an agent automatically checking for anything out of the ordinary without the user having to constantly approve every action (and getting notification fatigue in the process). “While these measures meaningfully reduce the risk, they cannot eliminate it.” — Anthropic. Thankfully, Claude will still ask the user before making any purchases or sharing personal data, though. You wouldn’t want to come home to a bunch of unexpected Amazon packages because Claude decided to buy itself some costly RAM. There is no perfect safety in agent land, though, so Anthropic rightly notes that “while these measures meaningfully reduce the risk, they cannot eliminate it.” Availability The new extension experience is now rolling out to users on the Max and Team plans. Pro users will see it in the coming weeks. For users on Enterprise plans, it is off by default, but admins can enable it and limit its use to specific domains. One more caveat: while there are plenty of other Chromium-based browsers, Anthropic notes that the extension won’t run on them (or on mobile). The post Anthropic’s Chrome extension is now a Cowork session appeared first on The New Stack.