Join CyberArk's security researchers as they unveil critical vulnerabilities in the Model Context Protocol (MCP) and demonstrate how organizations can protect themselves in the AI era. This session explores the emerging security challenges as AI applications become increasingly integrated into enterprise environments through standardized protocols like MCP.
Discover how the Model Context Protocol, designed as the "USB Type-C port for AI applications," creates new attack vectors through tool poisoning, command injection, and composability chaining. Learn about real-world attack scenarios where malicious MCP servers can hijack execution flows, steal sensitive data, and compromise endpoint security. Our researchers will demonstrate live examples of these vulnerabilities and show how attackers can exploit the trust relationships between MCP hosts, clients, and servers.
Explore comprehensive mitigation strategies using endpoint privilege management solutions to create security boundaries in the MCP ecosystem. See how application control, least privilege enforcement, and file access restrictions can protect your organization's AI infrastructure. Understand the importance of implementing proper security controls as your teams adopt new AI tools and protocols, ensuring they can innovate safely without exposing your organization to unnecessary risks.
