Malicious AI Skills in China Enable Jailbreaking and Illegal Crypto Mining

Thumbnail Image

The information displayed in the AIM should not be reported as representing the official views of the OECD or of its member countries.

China's National Internet Emergency Center (CNCERT) warned that some AI agent Skills are being used to bypass large language model safety restrictions and conduct illegal cryptocurrency mining. These malicious Skills have led to illegal content generation, account bans, device degradation, and users' involuntary involvement in criminal activities such as money laundering.[AI generated]

Why's our monitor labelling this an incident or hazard?

The announcement explicitly involves AI systems (large language models and their skill packages) and highlights direct and indirect harms caused by their malicious use, including violations of legal rights, harm to users' devices, and potential involvement in criminal activities. Although the harms are described as risks and potential consequences, the warning indicates that such harms have occurred or are occurring, making this an AI Incident rather than a mere hazard or complementary information. The AI system's misuse is pivotal in causing or enabling these harms.[AI generated]
AI principles
SafetyRobustness & digital security

Industries
Digital security

Affected stakeholders
Consumers

Harm types
Economic/PropertyReputational

AI system task:
Content generationGoal-driven organisation


Articles about this incident or hazard