HIGHLIGHTS
SUMMARY
Within the notion of cyberattacks, one can confidently consider adversarial AI attacks aimed at defrauding an AI system in such a way that the result is undetected by humans. Adversarial attacks against AI are a very modern form of fraud against underwriters; to understand them, in Section 2 the authors review frauds in the insurance sector and their impact. The authors associate some typical applications of AI with their corresponding example of adversarial attack/fraud. If the system was attacked by either PGD or by a universal attack, this time the tumor will be . . .
If you want to have access to all the content you need to log in!
Thanks :)
If you don't have an account, you can create one here.