TL;DR
NEAR AI contained unauthorized access after an attacker exploited an SGLang remote code execution vulnerability, reaching an inference server and internal infrastructure.
About 100 users had account details downloaded, 18 legacy agents exposed chat history and logs, and two customers had API keys exposed and revoked.
NEAR AI patched inference servers, rotated credentials and restored Cloud access, while reporting no evidence that Private Chat conversations or uploaded files were accessed.
NEAR AI contained unauthorized access to part of its infrastructure after an attacker exploited a remote code execution vulnerability in SGLang, the open-source inference engine running on its servers. According to the official incident report , the attacker gained code execution on one inference server before reaching internal systems. NEAR AI says the access has been closed, affected credentials replaced and the vulnerability patched across its inference servers. The incident affected services connected to NEAR Protocol . The report does not identify NEAR’s blockchain as affected.
NEAR AI Details Data Exposure and Response
The investigation confirmed that the attacker reached a system behind the deprecated Private Chat product and downloaded account details for approximately 100 users. Records included email addresses and names, plus plan and balance information for some accounts. NEAR AI found no indication that stored Private Chat conversations or uploaded files were accessed. The attacker also accessed 18 agents on a legacy NEAR AI Agents deployment outside the project’s confidential-computing TEE environment, exposing chat history, settings and logs.
On November 8, the party could also have observed user requests being processed on the compromised server, including requests running in confidential mode. NEAR AI explained that confidential computing protects data from operators and hosts but cannot prevent exploitation of vulnerable software inside the enclave. The SGLang flaw therefore bypassed protections at the software layer rather than breaking the TEE hardware model. The incident creates a security test for NEAR AI’s confidential inference infrastructure .
Two NEAR AI Cloud customers had API keys exposed in logs, and those keys were revoked. The team also revoked active sessions and administrative credentials, replaced agent tokens, took Private Chat offline, rotated credentials and completed a forensic review. Affected users were contacted directly, while NEAR AI said it would make required regulatory notifications. The response follows earlier security work across the ecosystem, including the NEAR Intents exploit recovery .
NEAR AI reported the vulnerability to SGLang and said it will apply stricter security analysis around its enclaves. It warned users about messages referencing accounts or balances. At drafting time, NEAR traded near $3.60 on CoinMarketCap, up 2.8% over 24 hours. The market snapshot shows no immediate token-price collapse tied to the disclosure, but it should not be treated as proof that the incident had no market impact. The episode arrives as NEAR expands AI-linked infrastructure .
