In a controlled laboratory environment, an uncensored, locally hosted artificial‑intelligence model was used to modify a Windows credential‑dumping utility, enabling it to bypass detection by two Endpoint Detection and Response (EDR) products.
The experiment was documented by Project Black researcher Eddie Zhang, who targeted the Local Security Authority Subsystem Service (LSASS) memory as the dump source, demonstrating the model’s capability to adapt offensive software.
ALSO READ | US Moves to Ban Chinese Advanced Robots Over Espionage Risks, Builds Own AI Infrastructure
Zhang’s findings underscore that readily accessible generative AI can accelerate the development of custom offensive tools, raising concerns for defenders about the speed at which such evasion techniques can be produced.