prompt injection

WebMCP Security Risks Put AI Agents Under Pressure

WebMCP security flaws are exposing AI agents to prompt injection, tool abuse, and data leaks as developers rush to connect agents to the web.