Perplexity BrowseSafe: Safer AI Browsers
Perplexity introduces BrowseSafe, an open detection model and benchmark for protecting AI agents from prompt injection attacks in browser environments.