# GCG Adversarial Suffix Attack

- **Severity:** High
- **Signature ID:** `f2d5647a-7d59-40a2-b1fa-ca454a739d02`
- **MITRE ATLAS:** AML.T0051 (LLM Prompt Injection)
- **OWASP:** ASI01 (Agent Authorization and Control Hijacking), LLM01 (Prompt Injection)

## Summary

Attackers can append a specific kind of computer-generated 'nonsense' text to a normal-looking request, which tricks an AI model into bypassing its safety training and doing something it would otherwise refuse. This detection looks for the telltale signature of that nonsense text.

## How the attack works

An attacker uses an automated technique (Greedy Coordinate Gradient, or GCG) to compute a short string of tokens that, when attached to a harmful request, statistically pushes the AI model toward complying instead of refusing. These suffixes look like garbage to a human: dense clusters of brackets, LaTeX or code fragments, mixed-language words, and odd punctuation. The attacker submits a normal-sounding prompt with this suffix tacked on the end. If the model is vulnerable, it produces the response the safety training was supposed to block.

## Why it matters

An organisation running an AI agent could have its safety filters bypassed, letting attackers extract restricted content, trigger disallowed actions, or manipulate agent behavior beyond intended guardrails.

## What you can do

- Log and review prompts containing dense bracket sequences, LaTeX-code hybrids, or mixed-language token salad, especially when paired with otherwise-blocked requests.
- Expect false positives from legitimate LaTeX documents, multilingual academic writing, and code reviews with nested array literals; check context before treating a match as an attack.
- Add input length and entropy checks ahead of the model to flag unusually high-entropy trailing text.
- Keep model safety training and filtering layered with independent output validation, since suffix attacks target the model directly and can bypass prompt-level defenses alone.

## Known benign look-alikes

- LaTeX source documents containing \usepackage{} and mathematical notation
- Multilingual academic text with German words followed by mathematical brackets
- Source code with deeply nested array literals (legitimate code review context)
- Repetitive !!! in ASCII art or emphasis sequences in informal messages

## References

- https://agentthreatrule.org/en/rules/ATR-2026-00267
- https://github.com/NVIDIA/garak/blob/main/garak/probes/gcg.py
- https://arxiv.org/abs/2307.15043

---
Source: https://www.netzilo.com/threats/atr-gcg-adversarial-suffix-attack