Analyzing the security risks of agentic browsing, specifically prompt injection via HTML, and exploring benchmarks like BrowseSafe.
BrowseSafe is a benchmark suite designed to evaluate these defenses. It tests agents against a dataset of diverse injection attacks embedded in realistic web pages.
The percentage of attacks that successfully manipulate the agent.
How often legitimate content is flagged as malicious.
The time added to the browsing session by the defense mechanism.