SGLang RCE Exploit Led to NEAR AI Data Access, Company Says
Summary
NEAR AI says an attacker exploited an RCE flaw in SGLang, accessed an inference server and internal systems, and downloaded details for about 100 users plus data from 18 agents. The company revoked exposed API keys, deployed a fix, and restored service.
Key points
- NEAR AI says an attacker exploited a remote code execution flaw in SGLang on one of its inference servers.
- Because SGLang ran inside a trusted execution environment, the attacker could have seen requests processed there, including confidential-mode requests; NEAR AI has not said whether any were accessed.
- The company confirmed the attacker downloaded account details for about 100 Private Chat users, including names and email addresses, and some plan and balance information.
- The attacker also accessed 18 agents on an older deployment outside the TEE and took chat history, settings, and logs.
- Two customers' NEAR AI Cloud API keys were exposed in logs; the company says it revoked them.
- NEAR AI revoked sessions and administrator credentials, replaced agent tokens, took Private Chat offline, deployed a fix, and rotated remaining credentials.
- NEAR AI says Cloud service has been restored; it did not identify a CVE or affected SGLang versions, or say whether the SGLang project has released a fix.
Article Details
- Victim Organization
- NEAR AI
- Victim Domain
- near.ai
- Incident Type
- Remote code execution exploit resulting in unauthorized system access and data theft
- Data Types Exposed
- User names and email addresses
- Some users' plan and balance information
- AI agent chat history, settings, and logs
- Two customers' NEAR AI Cloud API keys
- Affected Records
- Account details for about 100 users were downloaded; data from 18 AI agents was accessed; two customers' API keys were exposed in logs.
- Operational Impact
- Private Chat was taken offline. NEAR AI revoked active sessions and administrator credentials, replaced agent tokens, rotated remaining credentials, deployed a fix for the SGLang flaw on its inference servers, and completed a forensic review. Access to NEAR AI Cloud was restored.
- Ransom Or Extortion
- Unknown
- Claim Status
- confirmed
Vendors
Products
NEAR AI AgentsThe attacker also reached 18 AI agents on an older deployment of NEAR AI Agents that did not run inside the company's TEE. From those agents, the attacker took data including chat history, settings and logs.NEAR AI CloudTwo customers' NEAR AI Cloud API keys were exposed in logs. The company has revoked both.Private ChatThe attacker reached an internal system behind Private Chat, a product the company had already deprecated. So far, the company has confirmed that the attacker downloaded account details for about 100 users there. TheseSGLangAn 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.