# Unauthorized Cryptominer Deployment via SSH

- **Severity:** High
- **Signature ID:** `ad53e1d9-4355-47c9-8ffc-d3287910fbdf`
- **MITRE ATLAS:** AML.T0051 (LLM Prompt Injection), AML.T0057 (LLM Data Leakage)
- **OWASP:** ASI01 (Agent Authorization and Control Hijacking), ASI08 (Supply Chain Attacks on Agent Ecosystems), LLM01 (Prompt Injection), LLM02 (Sensitive Information Disclosure)

## Summary

An AI agent with SSH access can be instructed to read a host's SSH configuration, connect to a remote server, and install and run a cryptocurrency mining script on it. This turns the agent into a delivery mechanism for unauthorized resource-hijacking malware.

## How the attack works

A user or injected prompt tells the agent to read the local SSH config to find target hosts and credentials. The agent then uses those SSH details to connect to a remote server. Once connected, it downloads or writes a mining script (e.g., for Ethereum) and executes it on that machine. The result is compute resources on infrastructure the attacker doesn't own being consumed for mining, without any human explicitly authorizing the deployment.

## Why it matters

The organization whose server is hijacked loses compute capacity, pays higher cloud or power bills, and may face service degradation. It also indicates that an agent's SSH credentials and command execution can be abused to compromise systems the agent has reach into, beyond the machine it runs on.

## What you can do

- Restrict which hosts an agent's SSH credentials can reach, and avoid giving agents broad SSH key access by default.
- Require human approval before an agent executes remote SSH sessions or runs scripts on remote hosts.
- Monitor outbound SSH connections initiated by agent processes and flag unexpected new destinations.
- Watch for scripts referencing mining pools, wallet addresses, or known miner binaries being written or executed via automated sessions.

## Known benign look-alikes

- DevOps using ssh to deploy legitimate Ethereum node software
- Sysadmins describing detection of unauthorized crypto-mining

## References

- https://agentthreatrule.org/en/rules/ATR-2026-01838
- https://github.com/ai-safety-institute/AgentHarm

---
Source: https://www.netzilo.com/threats/atr-unauthorized-cryptominer-deployment-via-ssh