BrowseSafe is a specialized detection model to defend against evolving prompt injection attacks. It is designed specifically to spot and block malicious instructions hidden in web pages before they can impact AI browser agents.

→ View original post on X — @perplexity_ai,