AITechRadarabout 2 hours ago
AI agents tricked into fake memory poisoning
4 min read
Forcepoint X-Labs published a threat model for persistent memory poisoning. Hidden webpage text becomes a durable "fact" agents trust weeks later, demonstrated against ChatGPT, Gemini, Claude, and Microsoft 365 Copilot.
Level
Hype check
Tap to vote and see what everyone thinks.
