4 papers
Disentangled Safety Adapters Enable Efficient Guardrails and Flexible Inference-Time Alignment
Kundan Krishna, Joseph Y Cheng, Charles Maalouf +1
Existing paradigms for ensuring AI safety, such as guardrail models and alignment training, often compromise either inference efficiency or development flexibility. We introduce Di…
VLSU: Mapping the Limits of Joint Multimodal Understanding for AI Safety
Shruti Palaskar, Leon Gatys, Mona Abdelrahman +9
Safety evaluation of multimodal foundation models often treats vision and language inputs separately, missing risks from joint interpretation where benign content becomes harmful i…
SafetyPairs: Isolating Safety Critical Image Features with Counterfactual Image Generation
Alec Helbling, Shruti Palaskar, Kundan Krishna +3
What exactly makes a particular image unsafe? Systematically differentiating between benign and problematic images is a challenging problem, as subtle changes to an image, such as…
Learning to Reason for Hallucination Span Detection
Hsuan Su, Ting-Yao Hu, Hema Swetha Koppula +7
Large language models (LLMs) often generate hallucinations -- unsupported content that undermines reliability. While most prior works frame hallucination detection as a binary task…