TREE NEWS update: NEAR AI disclosed that an attacker exploited a remote code execution vulnerability in the open-source SGLang inference engine on one of its servers, downloading the account information of about 100 users, including emails, names and some plan and balance details. Eighteen NEAR AI Agents on legacy deployments outside a trusted execution environment were also accessed, exposing chat history, settings and logs. Two Cloud customers’ API keys found in logs have been revoked.
NEAR AI says attacker stole data of about 100 users via SGLang flaw
AI take
The exposure is small in headcount but notable in kind: an inference-engine flaw reached beyond model outputs into account and agent data, and the revoked API keys show how logging can quietly widen the blast radius. The detail that the affected agents ran on legacy deployments outside a trusted execution environment puts the spotlight on how AI infrastructure is hosted, not just which model is served. Whether the trusted-execution distinction becomes a broader segmentation standard across inference providers is the open question.
Generated by AI for reference only.
Share on WeChat
Open WeChat → Scan → then tap "…" to send to a chat or Moments.
Tap "…" in the top-right corner to send to a chat or share to Moments.
Related News
57m ago
UK Regulator: Some Banks Are Cutting AI Risk as Concerns Mount
1h ago
JD Technology Splits into Cloud and Fintech Units in Restructuring
1h ago
OKX launches ‘Contract Combo’ for multi-asset perpetual futures trading
1h ago
ETH falls below $2,100, down 1.27% in 24 hours
1h ago
Binance to Raise Collateral Ratios for FIL and HBAR on Nov 13
1h ago
UXLINK Hacker-Linked Address Closes 30,000 ETH Long at $1.1M Loss