Search papers, labs, and topics across Lattice.
Affiliation:
1
0
2
Small, human-imperceptible image perturbations can drastically mislead Vision Language Models, revealing a significant security flaw in multimodal AI systems.
