What type of attack is specifically aimed at disrupting AI-powered services?

What type of attack is specifically aimed at disrupting AI-powered services?

• A. Adversarial attacks
• B. Phishing
• C. Social engineering
• D. DDoS

Answer: (A)

Adversarial attacks target the AI system itself by feeding inputs crafted to make the model behave incorrectly. When an AI-powered service receives these specially designed inputs, it can produce wrong outputs, degrade performance, or fail in ways that disrupt the service. This direct manipulation of the model’s behavior is what makes adversarial attacks especially disruptive for AI-powered services. Phishing and social engineering aim at people, not the AI model, so they don’t directly disrupt how the AI system operates. DDoS tries to overwhelm a server to deny access, but it’s not about exploiting the AI model’s behavior itself; it’s about availability at the infrastructure level.

Adversarial attacks target the AI system itself by feeding inputs crafted to make the model behave incorrectly. When an AI-powered service receives these specially designed inputs, it can produce wrong outputs, degrade performance, or fail in ways that disrupt the service. This direct manipulation of the model’s behavior is what makes adversarial attacks especially disruptive for AI-powered services.

Phishing and social engineering aim at people, not the AI model, so they don’t directly disrupt how the AI system operates. DDoS tries to overwhelm a server to deny access, but it’s not about exploiting the AI model’s behavior itself; it’s about availability at the infrastructure level.