Memory Poisoning

Memory poisoning refers to attacks where malicious actors inject false or corrupted data into an AI system's training or operational memory to manipulate its outputs or behavior. This threat is particularly critical for enterprise governance because it can compromise model reliability, enable subtle decision-making errors, and create compliance violations without obvious detection. Organizations must implement data validation, access controls, and continuous monitoring to identify and prevent memory corruption that could undermine AI trustworthiness and accountability.

1 item

ResearchGlobal2026-08-18

Copilot Flaw Enabled One-Click Password Theft via Undocumented URL Parameter

Varonis researchers discovered a critical vulnerability in Microsoft 365 Copilot that allowed attackers to exfiltrate passwords and other sensitive data when a target clicked a single malicious link. The attack exploited an undocumented URL parameter that Copilot itself revealed through iterative questioning, bypassing the tool's consent guardrail. A separate but related vector enabled persistent memory poisoning that survived credential resets and session revocations.

Microsoft Copilot Microsoft prompt injection data exfiltration memory poisoning AI security enterprise AI Varonis