Question

An attacker fools an AI-based anti-spam system into thinking the attacker's email messages are benign. This enables the attacker's emails to reach their destination(s) without being blocked. What type of adversarial machine learning is this?

A.Spoofing
B.Poisoning
C.Evasion

Answer 1

Evasion is the type of adversarial machine learning where an attacker modifies input data to fool an AI system, allowing malicious emails to bypass anti-spam filters.

Step-by-step explanation:

When an attacker fools an AI-based anti-spam system into thinking that their email messages are benign, thereby enabling the messages to reach their destinations without being blocked, this type of adversarial machine learning is known as C. Evasion

Evasion attacks involve making subtle changes to the input data (in this case, the email content) that cause the machine learning model to make a mistake in its predictions. Hackers use these techniques, alongside other methods, to break into websites, get people to click on links, and lure individuals into giving out information.