inference servers

NEAR AI Breached Through an Inference Engine Flaw (Account Details of About 100 Users): What Was Confirmed and What Users and AI Infrastructure Operators Should Do

NEAR AI, which offers AI chat and agents, disclosed unauthorized access to part of its infrastructure in an official blog post dated November 9, 2026. According to the company, the cause was a remote code execution flaw in SGLang, the open-source inference engine running on its servers; the attacker ran code on one inference server and reached other infrastructure from there. What it confirmed: account details of about 100 users (email addresses, names, and for some accounts plan and balance information) taken from an internal system behind the previously deprecated Private Chat; chat history, settings and logs of 18 legacy agents that were not running in its confidential computing (TEE) environment; and, separately, NEAR AI Cloud API keys of two Cloud customers that were exposed in logs (revoked; the report does not say they were taken). On November 8 the attacker could also have observed user requests being processed on the compromised server, including confidential-mode requests. The company found no indication that stored Private Chat conversations or uploaded files were accessed. It says it has replaced credentials, fixed the flaw and restored access to NEAR AI Cloud (Private Chat remains offline). The official report gives no CVE number. Our view: for AI infrastructure operators, the key points are keeping inference servers apart from internal systems, leaving no legacy environments behind, and keeping secrets out of logs.