No Safe Dose: How Training Data Drives Unsafe Image Generation
cs.CV, cs.LG
Submitted: 2026-05-27
Updated: 2026-11-06
Code: https://github.com/Photoroom/PRX
Terminology
Sources
- Multimodal datasets: misogyny, pornography, and malignant stereotypes
- Beyond Overcorrection: Evaluating Diversity in T2I Models with DivBench
- Gemma 2: Improving Open Language Models at a Practical Size
- AILuminate: Introducing v1.0 of the AI Risk and Reliability Benchmark from MLCommons
- A Systematic Study of Bias Amplification
- Llama Guard: LLM-based Input-Output Safeguard for Human-AI Conversations
- Scaling Laws for Neural Language Models
- Aligning Text-to-Image Models using Human Feedback
- DiT-Air: Revisiting the Efficiency of Diffusion Model Architecture Design in Text to Image Generation
- The Cake that is Intelligence and Who Gets to Bake it: An AI Analogy and its Implications for Participation
- MSTS: A Multimodal Safety Test Suite for Vision-Language Models
- Qwen-Image Technical Report
- ShieldGemma 2: Robust and Tractable Image Content Moderation
- AI Risk Categorization Decoded (AIR 2024): From Government Regulations to Corporate Policies
Related papers
- TEA: Text Encoder Alignment for Robust Concept Erasure in Text-to-Image Models
- Do Egocentric Video-Language Models Capture Both Hand- and Object-Centric Cues?
- Aura: Consistent Multi-Subject Video Generation via VLM-Grounded Semantic Alignment
- PVRF: All-in-one Adverse Weather Removal via Prior-modulated and Velocity-constrained Rectified Flow
- DLWM: Diverse Latent World Models for Efficient Multimodal Reasoning
- Out of Sight, Still in Mind: Token Compression for Omni-LLMs