AI Relay Servers Mask Chinese Access to US Frontier Models
More than 80,000 AI relay servers are helping users in China hide their identities while accessing cutting-edge US AI models, likely to clone them.
Dark Reading reports that more than 80,000 AI relay servers are being used to help users in China hide their identities when they connect to advanced large language models in the US. The activity is probably aimed at cloning those models.
Relay servers sit between a user and the target service, forwarding requests so the original connection is difficult to trace. The large number of relays gives Chinese users an anonymous route to frontier AI systems hosted in the US.
The report says the likely goal is to clone the models. The setup lets Chinese users copy advanced LLMs while keeping their identities hidden.
More in Security & Privacy
Canada Probes IDScan After Breach of 153M Driver's Licenses
Privacy Commissioner opens investigation into IDScan.net's security practices and breach notifications after stolen ID scans of 153 million people appeared on the dark web.
Sweden Fines Miljödata $183,000 for Breach Affecting 2.2 Million
Sweden's privacy regulator penalized IT provider Miljödata for security failures linked to a breach that exposed 2.2 million people's data.
Rogue External MFA Providers Can Steal Passwords in Login Attacks
Researchers show that attackers with privileged access can register a malicious external MFA provider that harvests passwords during otherwise legitimate logins.
Rogue AI Incidents Push Safety Debate From Theory to Practice
A wave of misalignment reports is forcing AI labs, enterprises, and governments to confront real-world control and security gaps.