AgentGuard's point is: don't let AI's 'memory' make decisions for you; every sensitive operation must be verified again on the spot.

GoPlus Security discloses that in AgentGuard AI, an attack called "Historical Memory Injection" exploits long-term memory to make the AI treat preferences as authorization, inducing sensitive operations such as refunds and transfers. Protective measures include: requiring real-time confirmation for related operations, considering "habit/old behavior" as high risk, making long-term memory writes traceable, triggering secondary verification with ambiguous instructions, and ensuring long-term memory cannot replace real-time authorization. The AI memory system should be treated as a potential attack surface and audited with a security framework.