Prompt Injection Attacks Targeting MCP Tool Calls
Attackers exploit LLMs' inability to distinguish instructions from data in tool descriptions.
Camille Whitfield
Reporter
Camille Whitfield is a reporter at Agent Governance Review covering mcp security. Based in Amsterdam, Camille has written for Agent Governance Review since 2018.
1 story · Amsterdam
Attackers exploit LLMs' inability to distinguish instructions from data in tool descriptions.