A critical BadHost vulnerability is putting millions of AI agents, inference servers, and production applications at risk, and exploitation requires nothing more than a single malformed HTTP header. Tracked as CVE-2026-48710, the flaw was uncovered on January 27, 2026, by security researchers at X41 D-Sec during a source code audit sponsored by the Open Source Technology Improvement Fund (OSTIF). The vulnerability […]
The post BadHost Exploit Exposes Sensitive AI Agent Server Endpoints appeared first on Cyber Security News.