Check Point – AI Agent Security

Check Point AI Agent Security is a platform designed to discover, assess and protect agentic GenAI applications across the environments where they run. The product builds a continuously updated inventory of agents, their tools, and connected MCP servers, then generates holistic risk ratings mapped to industry frameworks like OWASP and MITRE ATLAS. At runtime, the integrated AI Guardrails layer enforces real-time protections against prompt injections and jailbreaks, data leakage and PII exposure, content violations, malicious links, and off-policy agent behavior — including tool allow/deny controls and custom threat policies. The solution is model‑agnostic (works with hosted providers, open-source, and custom models), supports 100+ languages, and screens structured and natural language text. Deployment options include enterprise SaaS and self‑hosted installations, with centralized policy management, analytics, logging, and SIEM integration. Its security intelligence platform continuously updates from public research, community signals and red‑teaming feeds, providing large and growing threat datasets for near‑zero‑day protections. The product offers both a full AI Agent Security tier (discovery, posture and runtime protection) and a standalone AI Guardrails runtime tier for embedding detection into existing GenAI applications.