Anthropic Launches Claude AI Agent for Chrome

AI startup Anthropic has launched a Chrome extension for its AI agent, Claude, based on the Claude model. The extension is initially available to a select group of 1,000 Max plan subscribers, with a waiting list open for others.

Key Features:

• Sidebar Interaction: Users can interact with Claude in a sidebar within Chrome.
• Context Retention: The agent retains context of all browser activities.
• Autonomous Actions: Claude can perform tasks autonomously, completing certain actions without user input.

Competitive Landscape

Browsers are emerging as a new battleground for AI labs, given their role as the primary gateway to the internet. Earlier, Perplexity launched its AI-based solution, Comet, and even proposed acquiring Chrome for $34.5 billion. OpenAI has also expressed interest in a similar deal and is preparing to introduce its own AI browser.

Security Concerns

Anthropic warned that AI agents with browser access pose new security risks. Brave team previously highlighted vulnerabilities in Comet, such as indirect prompt injection attacks, where hidden site code could trick the AI into executing malicious instructions.

Safety Measures

• Access Restrictions: Users can limit Claude’s access to specific sites (e.g., financial services, adult content, pirated material).
• Confirmation Requests: Claude requires user confirmation before performing high-risk actions like publishing, purchasing, or transmitting personal data.

Future Outlook

In August, Coinbase developers Kevia Leffew and Lincoln Murr predicted that autonomous agents would dominate Ethereum usage, highlighting their growing influence in blockchain and beyond.