NEAR AI breach via SGLang flaw exposes about 100 users' account details
NEAR AI says an attacker used a remote code execution bug in the open-source SGLang inference engine running on its servers. Roughly 100 users had emails, names and some plan and balance details downloaded, 18 agents on an older deployment had chat history and logs accessed, and two cloud customers' API keys were exposed and have since been revoked. The firm says the hole is patched, Private Chat is offline for now, and it saw no sign that stored private conversations or files were reached.
