Archive — US — techcrunch
Anthropic launches a Claude AI agent that lives in Chrome
Anthropic has launched a research preview of a browser-based AI agent, Claude for Chrome, which allows users to interact with the AI in a dedicated window while maintaining browsing context. Currently available to 1,000 subscribers on a paid plan, Claude can take actions in the browser following user permission. This release highlights a competitive landscape in AI-powered browsers, amid safety concerns and pending antitrust decisions involving Google.
Key points
- Anthropic announced Claude for Chrome, a browser-based AI agent shortly available to 1,000 paid subscribers.
- Claude can interactively chat with users and perform tasks in their browser with permission.
- The rise of AI browser agents is part of a broader competition among tech companies, including OpenAI and Perplexity.
- Anthropic emphasizes novel safety risks associated with AI agents in browsers and has implemented defenses against prompt injection attacks, reducing their success rate.
- Users have control over Claude's access to specific websites and can revoke permissions for high-risk actions.
Why it matters
- The development aligns with current trends in AI implementations as companies rapidly enhance browser integrations.
- The competition reflects a significant push in the tech industry, especially with Google facing an antitrust case and potential changes in the browser market.
- AI agents in browsers may greatly improve user productivity, suggesting a future where tasks are offloaded to AI more seamlessly.
The launch of Claude for Chrome showcases Anthropic's commitment to advancing browser-integrated AI, emphasizing the competitive environment in AI technology while addressing critical safety concerns, setting the stage for future developments in the AI landscape.