Abstract
Text-guided image generation models, such as DALL-E 2 and Stable Diffusion, have recently received much attention from academia and the general public. Provided with textual descriptions, these models are capable of generating high-quality images depicting various concepts and styles. However, such models are trained on large amounts of public data and implicitly learn relationships from their training data that are not immediately apparent. We demonstrate that common multimodal models implicitly learned cultural biases that can be triggered and injected into the generated images by simply replacing single characters in the textual description with visually similar non-Latin characters. These so-called homoglyph replacements enable malicious users or service providers to induce biases into the generated images and even render the whole generation process useless. We practically illustrate such attacks on DALL-E 2 and Stable Diffusion as text-guided image generation models and further show that CLIP also behaves similarly. Our results further indicate that text encoders trained on multilingual data provide a way to mitigate the effects of homoglyph replacements.
Abstract (translated)
URL
https://arxiv.org/abs/2209.08891
