Copilot response asserts 'no risk' for a documented CVE

Brandon Rodriguez@BMRodriguez

•updated Aug 25, 2026•0•0•5

Detects Microsoft Copilot assistant responses that explicitly claim a documented CVE is safe or poses no risk. This behavior is consistent with potential memory-driven prompt manipulation or disinformation, where an attacker has poisoned the assistant's memory to downplay known security vulnerabilities.