Anthropic has announced the launch of a beta version of its browser agent, Claude, which runs as an extension in Chrome, TechCrunch reports . The new tool is available to 1,000 users on its Max plan, which costs between $100 and $200 per month. There is a waiting list for others who want to join.
Claude for Chrome runs in a sidebar window of the browser, preserving the context of all user actions. The agent can be configured to perform specific tasks in the browser, including actions on behalf of the user. By default, Claude does not have access to financial services sites, adult content, and pirated resources. Before performing risky actions — such as publishing, purchasing, or transferring personal data — the agent asks for permission.
Anthropic notes that the goal of the pre-launch is to identify and eliminate new security risks. The company has already implemented protection against prompt injection attacks, which can occur through hidden code on web pages. According to Anthropic, the effectiveness of such attacks has been reduced from 23.6% to 11.2%.
The launch of Claude for Chrome comes amid intense competition among AI browser developers. Perplexity has already introduced its own browser, Comet, OpenAI is preparing a similar product, and Google is integrating Gemini into Chrome.
All this comes against the backdrop of an antitrust case against Google, where the court could force the company to sell Chrome. At the same time, Perplexity has said it is ready to buy Google Chrome for an amount exceeding its own valuation.