Living off the AI
An intruder who already runs code on a machine uses the AI tooling installed there — the coding agents, their API keys, local model servers — instead of bringing tools of its own. Its work then looks like ordinary activity from trusted programs, for the same reason living-off-the-land attacks use built-in system utilities. Detecting it means asking who started the agent and who is using its keys, not only what the agent did.
How MoorAI handles it: Four detectors: an AI coding CLI started by an unexpected parent program (alert, or block by policy), a decoy AI key being used (Critical), a local model server reachable without MoorAI, and an unknown program calling an AI provider API.
See also decoy credential · shadow AI · non-human identity
Related terms
-
Decoy credential
Defence & response
A credential that looks real but is never used by any legitimate process — a fake API key, cloud key or token placed where an intruder or a misled agent…
-
Shadow AI
Attack concepts
AI tools, agents, models and MCP servers in use without the security team's knowledge or approval: a developer's own coding agent, a chatbot fed…
-
Non-human identity
Agents & architecture
An identity used by software rather than a person — a service account, API key, OAuth token, certificate or workload identity — usually backed by a secret…
Attack concepts
The mechanisms the families above exploit, plus the agent-specific ones that have no chatbot equivalent. If you only read one entry on this page, read indirect prompt injection.