Researchers Uncover 10 In-the-Wild Prompt Injection Payloads Targeting AI Agents (infosecurity-magazine.com)

• 10 new prompt injection attacks on AI agents discovered
• Attack vector: indirect injection via external sources
• Forcepoint recommends improved security measures

"Forcepoint researchers have discovered 10 novel indirect prompt injection attacks targeting AI agents in real-world environments. The attacks exploit external data sources to inject malicious commands, posing significant security risks to enterprise AI deployments."