Malicious Data

Malicious data encompasses the intentional introduction of corrupted or deceptive information into datasets used to train or operate machine learning models, posing significant threats to model integrity and security. Current research focuses on understanding and mitigating various forms of malicious data, including backdoor attacks (where triggers cause unintended outputs), data poisoning (introducing corrupted training examples), and the misuse of combinations of seemingly safe models to achieve malicious goals. This research is crucial for developing robust and trustworthy AI systems, impacting fields ranging from cybersecurity and content moderation to healthcare and beyond, by improving the safety and reliability of machine learning applications.

Papers

July 1, 2024

Large Language Models Are Involuntary Truth-Tellers: Exploiting Fallacy Failure for Jailbreak Attacks
Yue Zhou, Henry Peng Zou, Barbara Di Eugenio, Yang Zhang
Large Language Model Language Model Jailbreak Attack Logical Fallacy Malicious Data

June 20, 2024

Adversaries Can Misuse Combinations of Safe Models
Erik Jones, Anca Dragan, Jacob Steinhardt
Artificial Intelligence System Strategic Manipulation Adversary Agent Model Safety Malicious Data

June 10, 2024

Chain-of-Scrutiny: Detecting Backdoor Attacks for Large Language Models
Xi Li, Yusen Zhang, Renze Lou, Chen Wu, Jiaqi Wang
Backdoor Attack Malicious Data

March 25, 2024

Generating Potent Poisons and Backdoors from Scratch with Guided Diffusion
Hossein Souri, Arpit Bansal, Hamid Kazemi, Liam Fowl, Aniruddha Saha, Jonas Geiping, Andrew Gordon Wilson, Rama Chellappa, Tom Goldstein, Micah Goldblum
Backdoor Attack Scratch Project Guided Diffusion Poisoning Efficiency Malicious Data

March 12, 2024

WannaLaugh: A Configurable Ransomware Emulator -- Learning to Mimic Malicious Storage Traces
Dionysios Diamantopoulos, Roman Pletka, Slavisa Sarafijanovic, A. L. Narasimha Reddy, Haris Pozidis
Ransomware Attack Ransomware Attack Modeling Technique Malicious Data Novel Mimic Algorithm Ransomware Detection

December 6, 2023

Feature Analysis of Encrypted Malicious Traffic
Anish Singh Shekhawat, Fabio Di Troia, Mark Stamp
Machine Learning Model Traffic Data Malware Attack Malicious Traffic Feature Mining Malicious Data

August 14, 2023

Machine Unlearning: Solutions and Challenges
Jie Xu, Zihan Wu, Cong Wang, Xiaohua Jia
Technical Challenge Solution Path Machine Unlearning Unlearning Framework Data Removal Malicious Data Adaptive Machine Learning

May 24, 2023

Malicious or Benign? Towards Effective Content Moderation for Children's Videos
Syed Hammad Ahmed, Muhammad Junaid Khan, H. M. Umer Qaisar, Gita Sukthankar
Gameplay Video Nine Year Old Child Content Moderation Malicious Content Benign Input Malicious Data

December 12, 2022

Fighting Malicious Media Data: A Survey on Tampering Detection and Deepfake Detection
Junke Wang, Zhenxin Li, Chao Zhang, Jingjing Chen, Zuxuan Wu, Larry S. Davis, Yu-Gang Jiang
Timely Survey Deepfake Detection Deepfake Technique Image Forgery Tamper Detection Malicious Data

November 3, 2022

Try to Avoid Attacks: A Federated Data Sanitization Defense for Healthcare IoMT Systems
Chong Chen, Ying Gao, Leyu Shi, Siquan Huang
New Attack Data Poisoning Attack Internet of Medical Thing Poisoned Data Malicious Data Federated Data Sanitization Defense

April 11, 2022

Narcissus: A Practical Clean-Label Backdoor Attack with Limited Information
Yi Zeng, Minzhou Pan, Hoang Anh Just, Lingjuan Lyu, Meikang Qiu, Ruoxi Jia
Backdoor Attack Backdoor Trigger Clean Label Backdoor Attack Limited Information Clean Label Attack Malicious Data