Anthropic launches the browser-based AI agent Claude for Chrome

Anthropic announced the launch of a preview version of the browser-based AI agent Claude. The new tool is available as a Chrome extension to a limited group of users — 1,000 subscribers of the Anthropic Max plan. For others, a waitlist is open.

After installing the extension, Claude opens in a separate window and retains context of everything happening in the browser. Users can also grant the agent permission to perform actions in the browser and take on certain tasks.

The browser is rapidly becoming a battleground for AI labs. Perplexity recently launched its own browser, Comet, with a built-in agent, and OpenAI is reportedly preparing to launch its own browser with similar features. Google has already integrated Gemini into Chrome.

Claude includes safety features: the probability of a successful "prompt injection" attack has been reduced from 23.6% to 11.2%. Users can also restrict Claude's access to specific websites. By default, resources offering financial services, adult content, and pirated material are blocked. The agent will request permission before performing "risky actions," such as posting, purchasing, or sharing personal data.

Source: TechCrunch