SGLang Remote Code Execution Exploit Compromises NEAR AI Inference Server, Exposes Data of ~100 Users and 18 Agents
An attacker exploited an RCE flaw in the open‑source SGLang inference engine, gaining code execution on a NEAR AI server and stealing data from roughly 100 users and 18 agents. The breach highlights the importance of continuous vulnerability‑management controls for audit readiness.
ADTP Breach Watch· November 10, 2026· The Hacker News
SeverityCritical
Type🐛 Vulnerability
ConfidenceHigh
ReportedNov 10, 2026
Technology & SaaSAI SaaS platformsCloud inference service providersVulnerability ExploitData Exfiltration
What happened
A remote‑code‑execution flaw in SGLang was used to run malicious code on a NEAR AI inference server, allowing the attacker to view confidential requests and extract data belonging to about 100 users and 18 internal agents.
Why it matters for trust and compliance
The incident demonstrates why organizations must maintain continuous, auditable evidence of vulnerability remediation for third‑party components.
Map vulnerability‑management processes to control objectives for continuous monitoring.
Collect patch‑deployment evidence to support audit readiness across frameworks.
Who is affected
AI SaaS platformsCloud inference service providers
Recommended actions
Run an inventory of all SGLang instances and verify they are patched to the latest release.
Integrate automated scanning for RCE‑type flaws into CI/CD pipelines and retain remediation evidence.
Regulations library
Every US state privacy law, on one map
24 states have a comprehensive privacy law in the library, alongside the EU, UK, US federal and global laws: 102 in all, each explained in plain words with its sources. Select a state to see what applies there.