Massive Network of AI Proxy Servers Used for Malicious Activities Uncovered

• •Over 10,000 AI proxy servers identified in China facilitating malicious activities.
• •Proxies are used for bypassing region restrictions and conducting model distillation attacks.
• •Tools like Claude Relay Service and sub2api are integral to this malicious ecosystem.

Security researchers from Team Cymru have identified over 10,000 proxy servers in China facilitating malicious AI activities. These servers, termed 'transfer stations,' are primarily used to bypass geographic restrictions and conduct model distillation attacks against frontier AI models. The infrastructure allows actors to obscure their identities while accessing and abusing AI services, including credential sharing and resale. Tools like Claude Relay Service and sub2api have been linked to this ecosystem, which is designed to violate terms of service of AI providers. The findings highlight a significant risk to intellectual property and compliance for AI model providers. Current exploitation status indicates active use of these proxies for malicious purposes. This situation raises concerns about the integrity of AI systems and the potential for widespread abuse.