List your company

Prompt Injection Defense on VPI-Bench

96.5ASR (Amazon)None
-3.8622.19548.2574.305Apr 14, 2026

Evaluation results

MethodLinks
NoneAgent Framework=Browser-use Agent (Gemini-2.0-pro)2026.0496.584.284.2
System PromptAgent Framework=Browser-use Agent (Gemini-2.0-pro)2026.0492.9885.9685.96
System PromptAgent Framework=Claude Agent (Sonnet-3.7)2026.0442.237.85.6
NoneAgent Framework=Claude Agent (Sonnet-3.7)2026.0431.736.716.7
Guard-gpt-4oAgent Framework=Browser-use Agent (Gemini-2.0-pro)2026.0422.815.821.1
Guard-gpt-4oAgent Framework=Claude Agent (Sonnet-3.7)2026.0410.612.24.4
WebAgentGuard-8BAgent Framework=Claude Agent (Sonnet-3.7)2026.041.700
WebAgentGuard-4BAgent Framework=Claude Agent (Sonnet-3.7)2026.0400.60
WebAgentGuard-8BAgent Framework=Browser-use Agent (Gemini-2.0-pro)2026.04000
WebAgentGuard-4BAgent Framework=Browser-use Agent (Gemini-2.0-pro)2026.0401.80