NEAR AI breached via SGLang RCE flaw, 18 agents and 100 users hit

What happened

According to The Hacker News, an attacker exploited a remote code execution flaw in SGLang, an open-source inference engine, and ran code on one of NEAR AI’s inference servers “last Sunday” (the article is dated 2026-11-10; the exact calendar date is not confirmed). The company’s incident report is the basis for the account.

Key points reported by The Hacker News:

• The attacker could also have seen requests processed on that server that day, including requests sent in confidential mode. That mode uses a trusted execution environment (TEE) meant to keep requests private even from the company. NEAR AI said the TEE “does not protect against a vulnerability in the software running inside the environment,” and SGLang runs inside it.
• The attacker moved further in and reached an internal system behind Private Chat, a product the company had already deprecated. NEAR AI has confirmed that account details for about 100 users were downloaded: email addresses and names, and for some accounts, plan and balance information. It found no sign that stored Private Chat conversations or uploaded files were accessed.
• The attacker also reached 18 AI agents on an older deployment of NEAR AI Agents that did not run inside the TEE, and took data including chat history, settings and logs.
• Two customers’ NEAR AI Cloud API keys were exposed in logs. The company has revoked both.

DataBreaches.net also reported the breach on 2026-11-10, with a headline saying it exposes user details and data from 18 AI agents. Only that headline was available to us, so every detail above comes from The Hacker News.

Response steps reported by The Hacker News: 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 of its inference servers, rotated its remaining credentials and completed a full forensic review. NEAR AI Cloud access has been restored. The company reported the flaw to the SGLang project.

Not confirmed in the sources: the CVE number, the affected SGLang versions, and whether the SGLang project has released a fix.

Am I affected?

• NEAR AI users: the company has contacted users whose account details or agents it confirmed were accessed, from security@near.ai, with steps that apply to them. In a follow-up post it said users who have not heard from it do not need to do anything (The Hacker News).
• Users whose requests passed through the affected server: NEAR AI has not said whether such people were among those contacted, so this is not confirmed.
• Brave Leo users: The Hacker News notes that, in an October 15 technical write-up, Brave said its Leo assistant runs chats with the GLM 5.3 Flash model inside a TEE operated by NEAR AI. NEAR AI has not said whether the affected server handled any of those chats, so impact there is not confirmed.
• Other SGLang operators: no affected versions have been named, so whether your deployment is vulnerable is not confirmed.

What to do now

1. If you use NEAR AI, check for an email from security@near.ai and follow the steps in it. Per the company, users who have not been contacted do not need to act.
2. As NEAR AI asks, be careful with messages about your account or balance that you did not ask for (The Hacker News).
3. If you run SGLang yourself, watch the SGLang project for an advisory, CVE or fixed version. None is confirmed in the sources, so we cannot name a version to upgrade to.
4. Check back for updates; this advisory will be revised if NEAR AI or the SGLang project publishes the CVE, affected versions or a fix.

Sources

• The Hacker News: SGLang RCE Exploit Breaches NEAR AI, Steals Data From 18 Agents and About 100 Users
• DataBreaches.net: Near AI Breach Exposes User Details and Data From 18 AI Agents