Skip to page content

AEGLIX MIND

Connected intelligence.Security research, connected

AEGLIX news · Global

SGLang Flaw Exploited to Breach NEAR AI Inference Server and Access User Data

Based on reporting from The Hacker News

Published by the source: Wed, 11 Nov 2026 00:49:39 +0530

AEGLIX News

Last Sunday, an attacker leveraged a remote code execution vulnerability in SGLang to execute code on one of NEAR AI's inference servers, the company confirmed. This intrusion potentially exposed requests processed on that server, including those submitted through confidential mode. Confidential mode is designed to process requests within a trusted execution environment that keeps them private even from NEAR AI, yet the company acknowledged the TEE cannot shield against flaws in the software operating inside it. Since SGLang runs within that TEE, the vulnerability provided the attacker a pathway to access the requests handled there.

Following the initial compromise, the attacker penetrated deeper into NEAR AI's infrastructure and accessed an internal system supporting Private Chat, a product the company had already deprecated. The intruder also reached 18 AI agents hosted on an older NEAR AI Agents deployment that operated outside the company's TEE. NEAR AI verified that the attacker downloaded account details for roughly 100 Private Chat users, capturing email addresses and names, and for certain accounts, plan and balance data. Additionally, two customers' NEAR AI Cloud API keys appeared in logs.

NEAR AI found no evidence that stored Private Chat conversations or uploaded files were accessed. However, the attacker exfiltrated data from the 18 AI agents, including chat history, settings, and logs. The company disclosed the vulnerability to the SGLang project but has not assigned a CVE identifier, specified the affected SGLang versions, or indicated whether a fix has been released.

After detecting the attack, NEAR AI revoked all active sessions and administrator credentials, replaced all agent tokens, took Private Chat offline, deployed a fix for the SGLang flaw on all inference servers, rotated remaining credentials and completed a full forensic review. The company has revoked both exposed NEAR AI Cloud API keys.

Report prepared 2026-11-10 · Based on reporting from The Hacker News. Source publication date above; subsequent developments may not be reflected.

Sources

Read original source · The Hacker News (opens in a new tab)

Privacy policy

Explore the story

Consult the cited publisher for the original reporting, updates and supporting evidence. Check the publication date and source before acting on a headline.

These are research tools; no CVE, indicator, or campaign relationship is inferred from this article.

More from The Hacker News

← Back to security news