Discovering Universal Semantic Triggers for Text-to-Image Synthesis
This work addresses ethical and robustness issues in text-to-image synthesis for AI safety and model auditing, though it is incremental as it builds on existing model vulnerabilities.
The paper tackles the vulnerability of text-to-image models by introducing Universal Semantic Triggers, meaningless token sequences that can manipulate generated images toward preset semantic targets, and finds that mainstream open-source models are susceptible to these triggers, posing ethical threats.
Recently text-to-image models have gained widespread attention in the community due to their controllable and high-quality generation ability. However, the robustness of such models and their potential ethical issues have not been fully explored. In this paper, we introduce Universal Semantic Trigger, a meaningless token sequence that can be added at any location within the input text yet can induce generated images towards a preset semantic target.To thoroughly investigate it, we propose Semantic Gradient-based Search (SGS) framework. SGS automatically discovers the potential universal semantic triggers based on the given semantic targets. Furthermore, we design evaluation metrics to comprehensively evaluate semantic shift of images caused by these triggers. And our empirical analyses reveal that the mainstream open-source text-to-image models are vulnerable to our triggers, which could pose significant ethical threats. Our work contributes to a further understanding of text-to-image synthesis and helps users to automatically auditing their models before deployment.