Claude now has its own browser and does not require the Chrome extension anymore.

Claude now has its own browser and does not require the Chrome extension anymore.

      Anthropic has integrated a browser directly into Claude, which opens in a side panel when a prompt requires web access, rather than using its Chrome extension. This feature is enabled by default for Enterprise, Pro, Max, and Team users, who have the option to switch back in the settings.

      The inclusion of a browser in Claude allows Cowork to recognize when a prompt needs the internet and displays a custom browser in a side panel, foregoing the need for Chrome through the Claude in Chrome extension. It is activated automatically. This browser is available for Enterprise plans and is being rolled out to Pro, Max, and Team users this week, with administrators or individual users able to revert to the extension in the settings.

      Anthropic’s messaging is notably straightforward. The company stated, “Many web tasks do not require your browser, just a browser, and now Claude has one.”

      The browser's design is intentionally narrow. It does not allow access to logins for email, banking, or other services unless entered manually, positioning it as a tool for retrieving public information rather than acting on behalf of the user.

      Chrome's dominance is significant; Statcounter reports that it accounts for nearly 70% of global web traffic. In Europe, there was a lengthy effort to open up this market. Article 6(3) of the Digital Markets Act mandates that gatekeepers present a choice screen upon first use, showing the primary browsers, search engines, and virtual assistants available, and it explains why the EU has compelled Google to allow competing assistants on Android.

      A browser embedded within an app bypasses these regulations. Anthropic is not classified as a gatekeeper, Claude is not considered a core platform service, and no choice screen appears for a browser that users never actively open.

      The security situation is transparent but not yet resolved. Anthropic acknowledges that prompt injection, where hidden instructions on a page can compromise an agent, remains a risk within the new browser, which utilizes the same protective measures as the Chrome extension.

      This security concern is not hypothetical for this product. TNW reported that Cowork could escape its local virtual machine and access credentials on a Mac.

      OpenAI has encountered similar issues. It shut down its standalone Atlas browser this month, integrating it into a Work mode that features its own built-in browser.

      Defaults are increasingly becoming the norm. Earlier, Anthropic also combined Claude chat and Cowork memory to be on by default, with the option to disable it later.

      Thus, Europe has legislated a screen that prompts users to choose a browser. However, the solution the industry is developing is one that users never see and were never asked about.

Other articles

Zoom features on smartphones are impressive. However, after trying the Vivo X300 Ultra, I realize they may not be suitable for everyone. Zoom features on smartphones are impressive. However, after trying the Vivo X300 Ultra, I realize they may not be suitable for everyone. I dedicated weeks to photographing with the Vivo X300 Ultra's 400mm ZEISS extender kit. The optics are exceptional, but the weight, setup time, and cost leave much to be desired. OpenAI, Anthropic, Google, and Microsoft advocate for cyber defense to be prioritized at the leadership level. OpenAI, Anthropic, Google, and Microsoft advocate for cyber defense to be prioritized at the leadership level. OpenAI, Anthropic, Google, and Microsoft are advocating for cyber defense to be prioritized at the leadership level. The EU has already implemented the reporting they are seeking. Anthropic is evaluating a new standard for Claude to interface with factory and laboratory equipment. Anthropic is evaluating a new standard for Claude to interface with factory and laboratory equipment. Anthropic's Model Hardware Standard allows vendors to define how an AI can control a machine. Starting in January 2027, Europe will oversee the safety features of AI in machinery. Private conversations with ChatGPT are increasingly being presented in legal proceedings. It’s important to exercise caution. Private conversations with ChatGPT are increasingly being presented in legal proceedings. It’s important to exercise caution. Conversations with chatbots can hold years of personal thoughts, intentions, and inquiries; however, they do not enjoy the legal privileges that safeguard conversations with attorneys. Daniel Dines, the founder of UiPath, has authored a book that challenges readers to disprove its claims by the year 2028. Daniel Dines, the founder of UiPath, has authored a book that challenges readers to disprove its claims by the year 2028. Daniel Dines anticipates that by the conclusion of 2028, no significant enterprise will conduct high-stakes tasks using a general agent, indicating that his framework is inadequate. Pollen Robotics' Microduck is designed to reduce the fragility and cost of training physical AI significantly. Pollen Robotics' Microduck is designed to reduce the fragility and cost of training physical AI significantly. Pollen Robotics' Microduck is a small biped priced at $399, created to facilitate physical AI training. It allows developers to test new robotic behaviors on a resilient platform that can withstand occasional falls.

Claude now has its own browser and does not require the Chrome extension anymore.

Anthropic has integrated a browser into Claude Cowork, which is enabled by default. The DMA choice screen is applicable only to gatekeepers, and Anthropic does not fall into that category.