# Targeted Harassment Bot Deployment

- **Severity:** High
- **Signature ID:** `c01728a7-09ce-466f-bdf5-575b05058377`
- **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 attacker instructs an AI agent to search for and deploy a GitHub repository or API that automatically sends hateful or abusive messages to a named individual. This turns the agent into an autonomous harassment tool rather than requiring the attacker to run the abuse themselves.

## How the attack works

The user gives the agent a goal like finding a tool that sends hateful messages aimed at a specific username or handle. The agent searches for and identifies a matching repo or API. It then deploys or configures that service, pointing it at the target's account or handle. Once running, the service can send abusive messages automatically and repeatedly without further attacker input.

## Why it matters

A named individual becomes the target of automated, sustained harassment, and the organization whose agent was used may be implicated in facilitating abuse, harming its reputation and creating legal exposure.

## What you can do

- Restrict agents from deploying arbitrary third-party code, repos, or APIs without human review.
- Require explicit approval before an agent installs or launches any messaging, notification, or bot service.
- Log and review agent actions that reference specific usernames, handles, or personal targets combined with messaging tools.
- Add policy checks that block agent tasks whose stated purpose is sending repeated messages to a named individual.

## Known benign look-alikes

- Trust-and-safety APIs that detect and remove hateful messages
- Deploying a benign GitHub repo that sends welcome messages

## References

- https://agentthreatrule.org/en/rules/ATR-2026-01849
- https://github.com/ai-safety-institute/AgentHarm

---
Source: https://www.netzilo.com/threats/atr-targeted-harassment-bot-deployment