Anthropic's Claude Cowork gains built-in browser to expand AI task execution

Anthropic introduces a built-in browser in Claude Cowork, transforming the AI’s role from information provider to active agent in web-based workflows, signalling a shift towards more practical AI applications.

Anthropic has added a built-in browser to Claude Cowork, widening the assistant’s role from information generation to task execution inside its desktop app. According to the company’s blog and support materials, the browser can open websites, read pages and fill forms directly, reducing dependence on a separate Chrome extension.

The key change is that Claude can now carry out multi-step web tasks within a side-panel browser while the user watches the process unfold. Anthropic says this is intended for work such as researching across sites, moving through pages, entering data and completing routine online workflows. In practical terms, the assistant is being positioned less as a question-answering tool and more as an agent that can perform parts of a job.

Anthropic has also stressed that the built-in browser is isolated from a user’s normal browsing environment. The company says it does not automatically inherit existing tabs, bookmarks or browsing history, which is meant to give users tighter control over what Claude can access. Its help centre also warns that browser agents should be used carefully on sensitive accounts and websites.

The new browser does not replace Claude in Chrome. Anthropic says the two tools serve different purposes: the built-in browser is suited to tasks Claude can complete independently, while the Chrome extension remains useful when the AI needs to work inside an already open session. For organisations, the company’s help pages say administrators can enable or disable browser use and manage settings for team and enterprise plans.

The update fits a wider trend in the AI sector, where competition is shifting from chat quality to practical execution. MacRumors reported in July that Claude Cowork had also expanded to mobile and web access, allowing users to monitor tasks across devices, while TechRadar described deeper Chrome integration as part of a broader push to make the assistant more useful across everyday workflows. The direction is clear: AI products are being built to do more than respond, and the central question is how much access users are willing to grant them.

Disclaimer: This content is intended for informational purposes only. Readers are advised to exercise their own judgement, conduct due diligence, or consult a qualified expert before acting on any information provided.