Work

On the Feasibility of Poisoning Text-to-Image AI Models via Adversarial Mislabeling

Year: 2025

Type: conference-paper

Abstract: Today's text-to-image generative models are trained on millions of images sourced from the Internet, each paired with a detailed caption produced by Vision-Language Models (VLMs). This part of the training pipeline is critical for supplying the models with large volumes of high-quality image-caption pairs during training. However, recent work suggests that VLMs are vulnerable to stealthy adversarial attacks, where adversarial perturbations are added to images to mislead the VLMs into producing incorrect captions.

Source: ACM SIGSAC Conference on Computer and Communications Security (CCS)

DOI: 10.1145/3719027.3744845

Authors: Stanley Wu, Ronik Bhaskar, Anna Yoo Jeong Ha, Shawn Shan, Haitao Zheng, Ben Y. Zhao

Institution: University of Chicago

Language: English

Topic: Adversarial Robustness in Machine Learning

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: hybrid

Funders: National Science Foundation, Office of Naval Research

Awards: G2642558705, Graduate Research Fellowship Program (GRFP), SaTC: CORE: Medium: Digital Forensics for Deep Neural Networks