
⚠️ HIGH — CVE-2026-100654 vLLM before 0.29.0 accepts user-controlled stop_token_ids on the OpenAI-compatible POST /v1/completions and POST /v1/ch… CVSS 6.5 ⚡ Exploit in the wild Full analysis → https://sec.kaitan.id/cves/CVE-2026-100654 #OpenAI #CyberSecurity #InfoSec
Post summary
The tweet warns of CVE-2026-100654 in vLLM, noting it is being actively exploited in the wild, with a CVSS score of 6.5 and a link to detailed analysis.
