NEAR AI breach: attacker used an SGLang flaw to grab about 100 users' account details and data from 18 AI agents

Published Tue, 10 Nov 2026 10:17:13 UTC · Updated Tue, 10 Nov 2026 23:15:00 UTC · Edited by Bizzle · AI-assisted

🔴Bearish for $NEARA security breach at the NEAR ecosystem's AI arm, which sells privacy as a core feature, is a reputational negative for NEAR.News sentiment, not financial advice. It reflects the likely impact of this news, not a price prediction.

Summary

NEAR AI, the AI product and infrastructure arm of the NEAR ecosystem, disclosed that an attacker exploited a remote code execution vulnerability in the open-source SGLang inference engine running on one of its servers, The Block and Decrypt reported, citing the company's incident report.

From there the attacker reached an internal system behind the already-retired Private Chat feature and downloaded names and email addresses for about 100 users, plus plan and balance information for some of them, according to both outlets.

The attacker also accessed chat histories, settings and logs from 18 agents on a legacy NEAR AI Agents deployment that was not running in the company's confidential computing environment, and logs exposed API keys belonging to two Cloud customers, which have been revoked.

NEAR AI said requests processed on the compromised server on Nov. 8, including confidential-mode ones, could have been observed, but it found no indication that stored Private Chat conversations or uploaded files were accessed; it has patched the flaw across all inference servers.

Key facts

NEAR AI: "Confidential computing protects data from the operator and the host. It does not protect against a vulnerability in the software running inside the environment, and that is what happened here," as quoted by The Block. (The Block)
The Block: NEAR AI Cloud access, paused during the investigation, has been restored, and the company revoked sessions and admin credentials, rotated agent tokens and took Private Chat offline. (The Block)
Decrypt: NEAR AI did not say whether the attacker used the two exposed API keys, and did not identify which SGLang flaw was exploited; it reported the vulnerability to SGLang's developers. (Decrypt)
Decrypt: NEAR AI said it has notified affected users, will make any required regulatory notifications, and warned users to be wary of unsolicited messages about their accounts or balances. (Decrypt)
The outlets differ slightly on the report date: The Block said it was published Monday, while Decrypt called it a Nov. 9 report. (Decrypt)