News
Anthropic Launches Claude for Chrome in Limited Beta, Security Risks Flagged
- By John K. Waters
- 08/26/2025
Anthropic has begun testing a Chrome browser extension that enables its Claude AI assistant to perform tasks directly within users' web browsers. The pilot, limited to 1,000 subscribers on the company's Max plan, marks Anthropic's first move into browser-based automation and is positioned as a research preview.
The extension enables Claude to perform actions such as scheduling meetings, filling forms, and navigating websites, effectively mimicking human interaction with web applications. However, internal testing highlighted vulnerabilities to "prompt injection" attacks, where hidden instructions embedded in websites or emails can trick the AI into harmful actions.
Anthropic said that red-teaming experiments showed these attacks succeeded nearly a quarter of the time before mitigations were introduced, and just over 11% after safeguards were implemented. Protections include site-level permissions, mandatory confirmations for sensitive actions, and blocked access to financial and adult sites.
The cautious rollout contrasts with more aggressive launches by rivals OpenAI and Microsoft, which have already introduced computer-controlling AI tools to broader user bases. Analysts see browser-based AI agents as a potential shift in enterprise automation, but one that carries ongoing security challenges.
About the Author
John K. Waters is the editor in chief of a number of Converge360.com sites, with a focus on high-end development, AI and future tech. He's been writing about cutting-edge technologies and culture of Silicon Valley for more than two decades, and he's written more than a dozen books. He also co-scripted the documentary film Silicon Valley: A 100 Year Renaissance, which aired on PBS. He can be reached at [email protected].