NEAR AI released a security incident report describing a remote-code-execution vulnerability in the SGLang component of its legacy deployment infrastructure. The issue was identified in a November 9 report and affected the AI-serving stack used to host language models.
Scope of the incident
The investigation found approximately 100 user details and 18 agents associated with the affected deployments. Two Cloud API keys were exposed and have been revoked. The company stated there is no evidence that stored private chat content was compromised, though it noted that some requests made on November 8 could potentially have been observed.
Technical background
SGLang is part of the service stack that routes requests, authenticates users, and logs activity for AI applications. A remote-code-execution flaw at this layer could allow an attacker to execute unauthorized instructions within the environment, depending on the privileges and network access available.
Response actions
NEAR AI reported that it investigated the affected deployments, revoked the exposed API keys, remediated the vulnerability, and notified impacted parties. The company emphasized that the incident relates to AI-serving infrastructure and does not affect the NEAR blockchain consensus rules or NEAR wallet keys.
Implications for crypto and AI security
The incident highlights the overlap between cryptocurrency and AI security, as crypto firms integrate AI tools into various workflows. It underscores the importance of isolating services, limiting credential exposure, monitoring access, and maintaining incident‑response plans for off‑chain components.
Recommendations for users
Users who may be affected should follow NEAR AI’s direct notifications and guidance, review any related credentials, and rotate keys where appropriate. There is no indication that ordinary NEAR wallet keys were exposed.
Next steps
NEAR AI is expected to publish a detailed technical post‑mortem outlining affected versions, root cause analysis, and validation of the fixes. The episode serves as a reminder that robust infrastructure security and transparent disclosure are essential for maintaining trust in AI‑driven financial services.
