THE SELF-EXPANDING STOLEN INFERENCE SUPPLY CHAIN: AN AI AGENT HARVESTING AND RE-SERVING LLM ACCESS

Summary from the TLDR InfoSec newsletter

An attacker's coding agent picked a SANS ISC honeypot as a free LLM backend and leaked 43 KB of its own control plane, including its offensive playbook, harvested API keys, and the operator's unproxied egress IP. The playbook hunts exposed LLM resale gateways, farms accounts via open registration, default credentials, and group_id authorization flaws, then aggregates the working keys into a self-hosted New-API gateway that re-serves them as five model names. Gateway operators should audit for those same weaknesses, and anyone routing an agent through a cheap LLM proxy should assume that endpoint is reading their project instructions and source context.