claim
Attackers may exploit vulnerabilities in proprietary foundational models to command the models themselves, potentially embedding malware or exploiting lesser-known attack surfaces in the model's feature space.

Authors

Sources

Referenced by nodes (1)