Railway breach investigator: LLM attacker's infrastructure still targeting other AI companies
cyb3rops · x · 2026-11-19
With Railway's permission, researcher eyalsela shared new details on the breach: the attacker's IP 70.36.107[.]74 has hosted "Ling" (a rebrand of ARTEX) and an "llm-api-key-checker" since mid-October, and the actor makes little effort to hide its activity.
Between Nov 10-15, Ling targeted at least two more companies: an AI platform provider (analyzing its Android app, writing Python clients for its private API, probing for SSRF against cloud metadata) and a startup providing runtime environments for AI agents (signing up with disposable mail[.]tm addresses).
Screenshots show Ling's web console, a bash loot script harvesting AWS IAM credentials and scanning env vars for keys containing openai/anthropic/gemini, and a port-scanning self-check.
Detection tip: LLM agents rely on mail[.]tm and similar disposable-email services (used in at least 11 attacks in the reported campaign), so monitoring those domains is a practical way to detect LLM-driven attacks.
More from coding & agent
- Devs flag GPT's defensive programming for hiding bugs; OpenAI asks for /feedback cases — pvncher · 2026-11-19
- Flutter ships A2UI 1.0: agents respond with real interactive widgets, not chat text — rseroter · 2026-11-19
- Skills vs MCP vs RAG vs Memory: the four context mechanisms AI agents actually need — MaryamMiradi · 2026-11-19
- Anthropic details agent memory: 'dreaming' lets Claude organize its experience — daniel_mac8 · 2026-11-19
- The case for externalized memory as the real fix for continual learning — daniel_mac8 · 2026-11-19
- FT: US tech outages blamed on rush to fix bugs found by AI — nordicinst · 2026-11-19