NEAR AI Contains Security Breach After SGLang Vulnerability Exploitation

NEAR AI runs inference infrastructure built on SGLang, an open-source engine, so a flaw in that layer can reach the systems that serve customer workloads. The November 9 incident report places this event in the context of a live service environment, not a theoretical test case. For builders using hosted AI services, the episode is a reminder that the security boundary includes the inference stack as well as the application layer.

For users and customers, the immediate issue is exposure of account details, chat history, logs, and API keys rather than a visible product feature change. Revoked keys mean integrations and automated workflows may need to be recreated or rotated before they work again. The restoration of cloud service suggests normal operations resumed, but the affected data and credentials still matter for anyone relying on the platform.

This case sits at the intersection of AI infrastructure and security hygiene, where a single remote code execution flaw can reach both service delivery and stored customer data. It also shows how quickly trust in managed AI tooling can depend on how operators isolate inference servers from internal systems. The trade-off is clear: centralized cloud AI services are convenient, but they also concentrate operational and security risk in one place.

Still unknown is the full scope of what the attacker accessed beyond the inference server, and whether any other data or systems were touched. The incident report described one path into internal infrastructure, but not every downstream step or the complete set of impacted customers. Further detail would come from a fuller postmortem, including the exact affected components and the remediation steps taken after the November 9 report.

NEAR ProtocolNEAR it is the asset of the company or protocol in the event

NEAR Protocol fell 1.1% in the 4 hours after the first headline, after rising 1.4% in the 2 hours before.

40 prices, one every 30 minutes · The move was larger before the first headline than after: an observation, not an explanation. The market also moves on things nobody publishes.