DiSCO: Defending text-to-image generation through distribution-guided contrastive prompt optimization
A new black-box defense mechanism effectively mitigates harmful outputs in text-to-image models via contrastive optimization.
As text-to-image generative models advance, they raise critical safety concerns, particularly the generation of Not-Safe-For-Work (NSFW) content such as violence and nudity, further exacerbated by red-teaming adversarial attacks. Existing defenses predominantly operate under whit...
ArXiv CS.AI
Source
Generated insight by OnAirToday AI
DiSCO provides a crucial safety layer for generative AI by intercepting prompts that trigger harmful content due to latent data distribution quirks rather than explicit text. This black-box approach offers a scalable solution for securing proprietary models where internal access is restricted.
Comments
Keep comments relevant and respectful. No links, advertising, self-promotion, or spam.