Prompt injection: when your AI agents are tricked into working for attackers
Companies are wiring AI agents into email, documents and internal systems. Hidden instructions in a web page or file can make an agent leak data.
Insights
Month: June 2026