
The information displayed in the AIM should not be reported as representing the official views of the OECD or of its member countries.
KAIST and international researchers developed 'ModelSpy,' a system that uses a small antenna to remotely extract AI model architectures by analyzing electromagnetic signals from GPUs, even through walls up to 6 meters away. This exposes a new physical security threat to AI assets, prompting proposed countermeasures.[AI generated]
Why's our monitor labelling this an incident or hazard?
The event involves the development and use of an AI system (AI models running on GPUs) and a novel attack method that can indirectly lead to harm by enabling theft of AI model designs. This constitutes a plausible future harm scenario (AI Hazard) because the attack could lead to intellectual property violations and compromise of critical AI infrastructure if exploited. Since no actual harm or incident has occurred yet, and the article focuses on the potential threat and mitigation strategies, the classification is AI Hazard.[AI generated]