AI RESEARCH

Consistent but Dangerous: Per-Sample Safety Classification Reveals False Reliability in Medical Vision-Language Models

arXiv CS.CV

ArXi:2603.20985v1 Announce Type: new Consistency under paraphrase, the property that semantically equivalent prompts yield identical predictions, is increasingly used as a proxy for reliability when deploying medical vision-language models (VLMs). We show this proxy is fundamentally flawed: a model can achieve perfect consistency by relying on text patterns rather than the input image. We