Which AI Security mentor is known for model vulnerabilities and attacks?

• A. Peter McCrory
• B. Sam Bowman
• C. Maxim Massenkoff
• D. Nicholas Carlini

Answer: (D)

Nicholas Carlini is the expert most associated with exploring model vulnerabilities and adversarial attacks in AI security. His work centers on how small, carefully crafted perturbations to input data can reliably cause machine learning models to misclassify, exposing weaknesses in defenses. The Carlini-Wagner attack, in particular, is a renowned optimization-based method that can bypass many defensive strategies by minimizing perturbations within strict constraints, underscoring the fragility of many models. This line of research has significantly shaped how researchers think about robustness and spurred the development of more resilient architectures and defenses. The other names are not as closely tied to adversarial manipulation and model robustness in the AI security literature, so they don’t fit this description as well.