As multimodal AI models advance from perception to reasoning, and even start acting autonomously, new attack surfaces emerge. These threats don’t just target…
As multimodal AI models advance from perception to reasoning, and even start acting autonomously, new attack surfaces emerge. These threats don’t just target inputs or outputs; they exploit how AI systems process, synthesize, and reason across modalities. The NVIDIA AI red team identifies and tests these vulnerabilities before attackers do. In our previous semantic prompt injection research…