← Back to article | Adversarial vulnerability assessment of vision language models for healthcare