FROM THE FRONTIER
Anthropic enters the AI browser wars with new Chrome extension
Click here to watch Anthropic’s new Chrome extension in action. Source: Anthropic
Anthropic just rolled out ‘Claude for Chrome’ — a new AI agent for Chrome, which adds its flagship Claude model directly into the browser. Dubbed an ‘early experiment’, it requires no extra logins, allowing the AI to read, summarize and act on web pages in real-time. You can see it in action here.
Unlike some other AI tools that operate in isolation, Claude for Chrome maintains full context of your browser activity through a sidecar window. Users on Anthropic’s Max plan ($100-200/month) can now delegate browser-based tasks directly to Claude, from research to form filling.
Browser agents are booming. The launch puts Anthropic squarely in the browser wars alongside Perplexity’s Comet browser and The Browser Company’s Dia, with OpenAI expected to enter the market as well. The timing is especially interesting given Google’s antitrust case, the result of which may force Google to sell Chrome.
But not all agents are created equal. A recent security report from browser maker Brave flagged serious security vulnerabilities in Perplexity’s Comet browser. Anthropic claims Claude for Chrome comes with built-in protections, reducing prompt injection attack success rates from 23.6% to 11.2%. The system also reportedly blocks access to financial, adult, and pirated content by default.
via Superhuman