Not Just Text: Uncovering Vision Modality Typographic Threats in Image Generation Models
This addresses security concerns for users and developers of image generation models by exposing overlooked vision modality vulnerabilities, though it is incremental as it builds on known attack methods.
The paper tackles the security risk of vision modality threats in image generation models, revealing that models are susceptible to typographic attacks and that existing defenses are ineffective, proposing a dataset for evaluating vulnerability.
Current image generation models can effortlessly produce high-quality, highly realistic images, but this also increases the risk of misuse. In various Text-to-Image or Image-to-Image tasks, attackers can generate a series of images containing inappropriate content by simply editing the language modality input. To mitigate this security concern, numerous guarding or defensive strategies have been proposed, with a particular emphasis on safeguarding language modality. However, in practical applications, threats in the vision modality, particularly in tasks involving the editing of real-world images, present heightened security risks as they can easily infringe upon the rights of the image owner. Therefore, this paper employs a method named typographic attack to reveal that various image generation models are also susceptible to threats within the vision modality. Furthermore, we also evaluate the defense performance of various existing methods when facing threats in the vision modality and uncover their ineffectiveness. Finally, we propose the Vision Modal Threats in Image Generation Models (VMT-IGMs) dataset, which would serve as a baseline for evaluating the vision modality vulnerability of various image generation models.