Claude is being tested as a Chrome extension to enhance browser-based AI capabilities while addressing security risks like prompt injection. The pilot aims to gather feedback on safety and usability before a broader release, with participants having control over what Claude can do and access.

This article outlines Google's advancements in Chrome's security, specifically addressing the risks associated with agentic browsing. It details measures like the User Alignment Critic, origin gating, and user confirmations to combat threats like indirect prompt injection and unauthorized actions. The goal is to ensure user safety while interacting with AI-driven features.

Google is enhancing Chrome with AI to create a smarter browsing experience that assists users in being more productive and secure online. The new features include an AI browsing assistant named Gemini, smarter search capabilities in the omnibox, and advanced safety measures to protect users from scams and privacy issues. These improvements aim to transform Chrome into a proactive partner that understands user needs and enhances web navigation.