all InfoSec news
Mudjacking: Patching Backdoor Vulnerabilities in Foundation Models
Feb. 26, 2024, 5:11 a.m. | Hongbin Liu, Michael K. Reiter, Neil Zhenqiang Gong
cs.CR updates on arXiv.org arxiv.org
Abstract: Foundation model has become the backbone of the AI ecosystem. In particular, a foundation model can be used as a general-purpose feature extractor to build various downstream classifiers. However, foundation models are vulnerable to backdoor attacks and a backdoored foundation model is a single-point-of-failure of the AI ecosystem, e.g., multiple downstream classifiers inherit the backdoor vulnerabilities simultaneously. In this work, we propose Mudjacking, the first method to patch foundation models to remove backdoors. Specifically, given …
arxiv attacks backdoor backdoor attacks build can cs.cr cs.cv cs.lg ecosystem failure feature foundation foundation models general patching point purpose single vulnerabilities vulnerable
More from arxiv.org / cs.CR updates on arXiv.org
Jobs in InfoSec / Cybersecurity
Product Regulatory Compliance Specialist
@ Avery Dennison | Oegstgeest, Netherlands
Cyber Security Analyst
@ FinClear | Melbourne, Australia
Senior Application Security Manager, United States-(Virtual)
@ Stanley Black & Decker | New Britain CT USA - 1000 Stanley Dr
Vice President - Information Security Management - FedRAMP
@ JPMorgan Chase & Co. | Chicago, IL, United States
Vice President, Threat Intelligence & AI
@ Arctic Wolf | Remote - Minnesota
Cybersecurity Analyst
@ Resource Management Concepts, Inc. | Dahlgren, Virginia, United States