DiSCO defends text-to-image models against NSFW attacks
DiSCO uses distribution-guided contrastive prompt optimization to defend text-to-image models against NSFW content generation and red-teaming attacks. The paper is on arXiv (2608.17067).
1 source
Daily brief
Get tomorrow's AI brief in your inbox
More stories today
- Max Hodak: Intelligence may be a law of physics
- User generates 100 portraits of Muhammad with ChatGPT
- User shares trick: ChatGPT creates custom podcasts for car rides
- Hugging Face CEO: Most AI workloads will run on open models
- Enterprises winning with AI agents are limiting agent autonomy