NEAR AI restores Cloud after security breach
4.1 Security NEAR AI suffered a breach via SGLang, exposing data of 100 users and 18 legacy AI agents. The flaw was patched, API keys revoked, and Cloud service restored. NEAR Protocol was not affected.
NEAR AI experienced a security breach beginning November 8, 2026, when an attacker exploited a remote code execution vulnerability in SGLang, the open-source inference engine used on its servers. The attacker accessed one inference server and reached internal systems, exposing account details—including email addresses and names—for about 100 users. Additionally, chat histories, settings, and logs from 18 legacy AI agents were compromised. Two customers had their Cloud API keys exposed, which were immediately revoked. In response, NEAR AI suspended its Cloud inference service, patched the vulnerability, rotated credentials, and conducted a forensic review. The company emphasized that confidential computing protected data from the operator and host, but not from vulnerabilities within the software itself. NEAR Protocol and other NEAR services were not affected by the incident. The Cloud service has since been restored, and NEAR AI reported no evidence that stored Private Chat conversations or uploaded files were accessed during the breach.
