Safe Text-to-Image Generation: Simply Sanitize the Prompt Embedding

Documenti analoghi

• BELT: Old-School Backdoor Attacks can Evade the State-of-the-Art Defense with Backdoor Exclusivity Lifting
  di: Qiu, Huming, et al.
  Pubblicazione: (2023)
• PIIGuard: Mitigating PII Harvesting under Adversarial Sanitization
  di: Liu, Mingshuo, et al.
  Pubblicazione: (2026)
• SafeGen: Mitigating Sexually Explicit Content Generation in Text-to-Image Models
  di: Li, Xinfeng, et al.
  Pubblicazione: (2024)
• PISanitizer: Preventing Prompt Injection to Long-Context LLMs via Prompt Sanitization
  di: Geng, Runpeng, et al.
  Pubblicazione: (2025)
• Finding a Wolf in Sheep's Clothing: Combating Adversarial Text-To-Image Prompts with Text Summarization
  di: Cooper, Portia, et al.
  Pubblicazione: (2024)