Skip to content
KitploitKITPLOIT
ToolsExploitsBlog
Log in
Submit
ToolsExploitsBlog
Submit

Hacking, PenTest, and Cybersecurity Tools for Your Security Arsenal!

Kitploit is a directory of hacking, cybersecurity, and pentesting tools. Discover the latest project updates to find vulnerabilities, analyze systems, automate testing, and strengthen your security.

FeedsContactPrivacy© 2026 Kitploit

Tool Directory

Categories

View all categories
Loading categories
Visco-Attack — Research pipeline implementing a visual-contextual jailbreak against multimodal LLMs: T2I prompt extraction, SD3.5 image rendering, multi-round spoof context construction, and GPT-4o-judged ASR evaluation. | Kitploit
Tools/GitHubGitHub/dtc7w3pq/visco-attack
Machine LearningPapers & ResearchRed TeamingAI SecurityAdversarial Attack
GitHubdtc7w3pq/visco-attack

Visco-Attack

Research pipeline implementing a visual-contextual jailbreak against multimodal LLMs: T2I prompt extraction, SD3.5 image rendering, multi-round spoof context construction, and GPT-4o-judged ASR evaluation.

View Repository
3117 months agoReviewed by Kitploit
Share

Visual Contextual Attack: Jailbreaking MLLMs with Image-Driven Context Injection

Paper Data HF Downloads GitHub Stars

VisCo-Attack Original pioneers a vision-centric attack paradigm and leverages multiple images across multi-turn dialogues; its tailored context-fabrication strategies deliver a notably high attack-success rate against leading MLLMs, positioning it as one of the strongest jailbreak methods for MLLMs to date.

If you are interested in our work, please star ⭐ our project, Thx 💕

⚠️ WARNING: This repo contains adversarial prompt techniques intended for research and safety purposes only.

📝 Overview

VisCo-Attack is our visual-contextual jailbreak framework introduced in the paper “Visual Contextual Attack: Jailbreaking MLLMs with Image-Driven Context Injection.” Instead of treating an image as a mere trigger, VisCo-Attack constructs a realistic, fabricated multi-turn dialogue context in which the visual input is essential to the harmful scenario. It automatically weaves that context with four image-centric strategies—Image-Grounded Scenario Simulation (VS), Multi-Perspective Analysis (VM), Iterative Interrogation (VI), and Hallucination Exploitation (VH)—then performs toxicity obfuscation and semantic refinement to yield a single attack prompt. Under a fully black-box setting, this prompt reliably breaks state-of-the-art MLLMs such as GPT-4o, Gemini 2.0 flash, and InternVL-2.5, pushing the attack-success rate to 85 – 91 % on MM-SafetyBench, FigStep-SafeBench, and HarmBench while maintaining toxicity scores above 4.7 / 5.

Visco-Attack Method Illustration

Figure: Overview of the Visual Contextual Attack (Visco-Attack) Method.

🚀 News

  • [2025/7/3] Release the attack contexts for four distinct strategies on MM-SafetyBench, now available on Hugging Face. Code will be released upon paper acceptance.
  • [2025/7/4] Our paper is now available on arXiv: https://arxiv.org/abs/2507.02844.
  • [2025/7/19] Provided SD + Typography images for each realigned MM-SafetyBench sample (SD image on top, keyword below). Used only for baseline reproduction; not part of our attack method.
  • [2025/8/14] Our Hugging Face dataset downloads have surpassed 10,000 🎉🎉🎉! Thanks for your support!
  • [2025/8/21] VisCo Attack was accepted to EMNLP 2025 Main Conference! The code will be released soon~
  • [Code release] The end-to-end pipeline (T2I prompt extraction → SD3.5 → 4-strategy multi-round spoof construction → target attack → GPT-4o judge) is now in this repo. See Quick Start below.

⚡ Quick Start

# 1. Install
pip install -r requirements.txt

# 2. Configure endpoints
#    Edit configs/default.yaml — fill in vLLM base_urls for the four roles:
#      models.generator   data-generation LLM
#      models.wizard      Wizard-Vicuna-Uncensored (or any unaligned LLM)
#      models.target      the target MLLM you want to attack
#      models.judge       judge LLM (paper used GPT-4o)
#    Smoke test: pointing all four at the same endpoint works fine.

# 3. Build a question file (or use the 3-question example)
python data/extract_questions.py \
    --input  /path/to/your/MMSafetyBench.json \
    --output data/mmsafebench_questions.json

# 4. Run the full pipeline
./run.sh

# Or run a single stage at a time:
RUN_STAGE=1 ./run.sh   # extract T2I prompts
RUN_STAGE=2 ./run.sh   # render core images (uses torchrun + SD3.5)
RUN_STAGE=3 ./run.sh   # multi-round spoof context (4 strategies × N gens × 4 rounds)
RUN_STAGE=4 ./run.sh   # render any extra images requested by stage3
RUN_STAGE=5 ./run.sh   # attack the target MLLM
RUN_STAGE=6 ./run.sh   # GPT-4o-as-judge ASR

🤖 Recommended red-team models (for generator / wizard roles)

The pipeline has 4 LLM roles in configs/default.yaml: generator, wizard, target, judge. The generator and wizard roles do the actual red-team work (constructing spoof contexts, producing fake-success continuations). They must be willing to follow the templated red-team instructions in prompts/*.md — safety-aligned models will typically refuse and the pipeline will degrade to empty / unparseable outputs.

The following models worked well for the red-team roles:

ModelRole(s) it suitsNotes
qwen-3.5-122B-uncensored-stxtgenerator + wizardA community Qwen3-derivative. vllm_client.py auto-disables Qwen3 thinking mode when it sees this model name.
Qwen2.5-72B-InstructgeneratorAligned, but typically still cooperates in producing fabricated unsafe contexts.

Models that frequently refuse (don't use them as generator or wizard):

  • GPT-4 / GPT-4o, Claude 3.5+ — strong alignment, will respond to the templated red-team prompts with refusals.
  • Llama-3-Instruct without further bypass.

Smoke test (3 questions, ~5 minutes)

If you'd rather verify the whole pipeline end-to-end before doing a real run, tools/run_real_smoke.sh is a one-shot wrapper that writes a tiny config (3 questions × 1 strategy × 1 gen × 2 rounds), runs all 6 stages with real SD3.5 diffusion, and prints per-stage file counts + ASR at the end. All paths are overridable via env vars:

SD_MODEL_PATH=/abs/path/to/stable-diffusion-3.5-large-turbo \
VLLM_URL=http://my-vllm-host:8000/v1 \
NUM_GPUS=N \
bash tools/run_real_smoke.sh

Bump INPUT_QUESTIONS / STRATEGIES / GEN_PER_QUESTION / ROUNDS env vars to scale up to the full paper-faithful run.

🗺️ Pipeline overview

Download Tool

Most Popular

View all →

Discover the most used tools by our community.

  • 1

    OsintHunter

    letox4/osinthunter

    🎯 The ultimate OSINT toolkit: investigation graph, confidence scoring, alias families, map, timeline and reports. 49 modules, 400+ sites, no API keys required.

    708
  • 2

    ghostlock-app

    yukonga/ghostlock-app

    GhostLock One-Tap Execution App (CVE-2026-43499)

    500
  • 3

    ARTEX

    baarai/artex

    LLM multi-agent autonomous penetration testing platform with planner/worker agents, asset and exploration graphs, MITM traffic recording, and vulnerability retesting.

    398
  • 4

    redforge

    cybersecurity91/redforge

    Browser-side red team toolkit with 102 offline tools for recon, payload generation, credential attacks, crypto, encoding and engagement reporting.

    341
  • 5

    CyberStrikeAI

    ed1s0nz/cyberstrikeai

    The system of action for AI-native cybersecurity—where intent becomes governed execution, evidence becomes operational memory, and every operation improves the next.

    259
  • 6

    antifox

    antifox-ai/antifox

    AI multi-agent autonomous penetration testing platform that runs recon, attack-surface discovery, exploitation validation and evidence-backed report delivery.

    257

Explore all tools

Browse our collection of tools

View all tools →