Why this issues for AI infrastructure
The weak inference servers kind the spine of many enterprise-grade AI stacks, processing delicate prompts, mannequin weights, and buyer information. Oligo reported figuring out 1000’s of uncovered ZeroMQ sockets on the general public web, some tied to those inference clusters.
If exploited, an attacker may execute arbitrary code on GPU clusters, escalate privileges, exfiltrate mannequin or buyer information, or set up GPU miners, turning an AI infrastructure asset right into a legal responsibility.
SGLang has been adopted by a number of massive enterprises, together with xAI, AMD, Nvidia, Intel, LinkedIn, Cursor, Oracle Cloud, and Google Cloud, Lumelsky famous.
