jiaxiaojunQAQ/FOA-Attack

Adversarial Attacks against Closed-Source MLLMs via Feature Optimal Alignment (NeurIPS 2025)

stars

commits

Python

primary language

Nov 5, 2025

updated

README

⚔️ Adversarial Attacks against Closed-Source MLLMs via Feature Optimal Alignment

FOA-Attack is proposed to enhance adversarial transferability in multimodal large language models by optimizing both global and local feature alignments using cosine similarity and optimal transport.

💥 News

[2025-09-19] Our paper has been accepted to NeurIPS 2025! 🎉
[2025-05-29] We release the FOA-Attack code! 🚀

💻 Requirements

Dependencies: To install requirements:

pip install -r requirements.txt

🛰️ Quick Start

python generate_adversarial_samples_foa_attack.py
python blackbox_text_generation.py -m blackbox.model_name=gpt4o,claude,gemini
python gpt_evaluate.py -m blackbox.model_name=gpt4o,claude,gemini
python keyword_matching_gpt.py -m blackbox.model_name=gpt4o,claude,gemini

⚙️ Start (automatic selection of cluster centers)

python FOAttack.py

💖 Acknowledgements

This project is built on M-Attack. We sincerely thank them for their outstanding work.

Contributors

jiaxiaojunQAQ

8 commits

Similar repos

VILA-Lab/M-Attack

[NeurIPS25 & ICML25 Workshop on Reliable and Responsible Foundation Models] A Simple Baseline…

102

alexoh2bd/AttackOnEmbeddings

experiments evaluating adversarial PGD attacks on multimodal embedding models

llm-attacks/llm-attacks

Universal and Transferable Attacks on Aligned Language Models

4,786

kaiyuanCui/UltraBreak

[ICLR2026] Toward Universal and Transferable Jailbreak Attacks on Vision-Language Models

saadiftikhar04/dlm-jailbreak-transfer

qroa/QROA

QROA: A Black-Box Query-Response Optimization Attack on LLMs

submission993/qroa

GFNOrg/red-teaming

Code for "Learning Diverse Attacks on Large Language Models for Robust Red-Teaming and Safety…

Languages

Python

100.0%

jiaxiaojunQAQ/FOA-Attack

Adversarial Attacks against Closed-Source MLLMs via Feature Optimal Alignment (NeurIPS 2025)

stars

commits

Python

primary language

Nov 5, 2025

updated

README

⚔️ Adversarial Attacks against Closed-Source MLLMs via Feature Optimal Alignment

FOA-Attack is proposed to enhance adversarial transferability in multimodal large language models by optimizing both global and local feature alignments using cosine similarity and optimal transport.

💥 News

[2025-09-19] Our paper has been accepted to NeurIPS 2025! 🎉
[2025-05-29] We release the FOA-Attack code! 🚀

💻 Requirements

Dependencies: To install requirements:

pip install -r requirements.txt

🛰️ Quick Start

python generate_adversarial_samples_foa_attack.py
python blackbox_text_generation.py -m blackbox.model_name=gpt4o,claude,gemini
python gpt_evaluate.py -m blackbox.model_name=gpt4o,claude,gemini
python keyword_matching_gpt.py -m blackbox.model_name=gpt4o,claude,gemini

⚙️ Start (automatic selection of cluster centers)

python FOAttack.py

💖 Acknowledgements

This project is built on M-Attack. We sincerely thank them for their outstanding work.

Similar repos

VILA-Lab/M-Attack

[NeurIPS25 & ICML25 Workshop on Reliable and Responsible Foundation Models] A Simple Baseline…

102

alexoh2bd/AttackOnEmbeddings

experiments evaluating adversarial PGD attacks on multimodal embedding models

llm-attacks/llm-attacks

Universal and Transferable Attacks on Aligned Language Models

4,786

kaiyuanCui/UltraBreak

[ICLR2026] Toward Universal and Transferable Jailbreak Attacks on Vision-Language Models

saadiftikhar04/dlm-jailbreak-transfer

qroa/QROA

QROA: A Black-Box Query-Response Optimization Attack on LLMs

submission993/qroa

GFNOrg/red-teaming

Code for "Learning Diverse Attacks on Large Language Models for Robust Red-Teaming and Safety…

Contributors

jiaxiaojunQAQ

8 commits

Languages

Python

100.0%