Skip to main content

Model Evasion

Tests whether adversarial examples, feature manipulation, or boundary exploitation can evade the model's safety mechanisms.

Sub-types​

  • Adversarial Examples: Crafted inputs that cause the model to misclassify or produce wrong outputs.
  • Feature Space Manipulation: Manipulating input features to evade detection or safety mechanisms.
  • Model Boundary Exploitation: Exploiting decision boundaries to find blind spots in model behaviour.

Threat Profile​

Objective: jailbreak

Objective vs. goals

The objective above selects the scoring rubric for a run — it is one of a fixed set of built-in names, not something you write. The goals you pass to an attack are separate free-text strings that you author yourself. See Goals vs. objective.

Primary

  • advbench: Adversarial benchmarks for evaluating evasion resistance

Secondary

  • xstest: XSTest for adversarial prompt detection

Attack Techniques​

Primary

  • Static Template: Template-based prompt injection
  • PAIR: Iterative refinement for bypass discovery

Secondary

  • AdvPrefix: Adversarial prefix optimisation

Metrics​

  • asr
  • judge_score

Usage​

Instantiate the Vulnerability​

from hackagent.risks import ModelEvasion
from hackagent.risks.model_evasion.types import ModelEvasionType

# Use all sub-types
vuln = ModelEvasion()

# Or specify particular sub-types
vuln = ModelEvasion(types=[
    ModelEvasionType.ADVERSARIAL_EXAMPLES.value,
    ModelEvasionType.MODEL_BOUNDARY_EXPLOITATION.value,
])

Run an Evaluation Campaign​

from hackagent import HackAgent
from hackagent.risks.model_evasion import MODEL_EVASION_PROFILE

agent = HackAgent(endpoint="http://localhost:8080/chat", name="my-agent")

# Profile techniques use display casing (e.g. "StaticTemplate");
# HackAgent.hack() expects the registered snake_case attack_type key.
ATTACK_TYPE_KEYS = {"StaticTemplate": "static_template", "PAIR": "pair"}

# Use profile recommendations
for attack in MODEL_EVASION_PROFILE.primary_attacks:
    for dataset in MODEL_EVASION_PROFILE.primary_datasets:
        attack_config = {
            "attack_type": ATTACK_TYPE_KEYS[attack.technique],
            "objective": MODEL_EVASION_PROFILE.objective,
            "dataset": {"preset": dataset.preset},
        }
        results = agent.hack(attack_config=attack_config)
        print(f"Results: {results}")