What is a potential mitigation for denial of service attacks in AI?

• A. Implementing rate limiting and anomaly detection
• B. Increasing data leakage controls
• C. Reducing system monitoring
• D. Limiting access to authorized users only during peak times

Answer: (A)

Denial of service attacks aim to make an AI service unavailable by overwhelming its resources. The best mitigation combines preventing overload with spotting abnormal traffic. Rate limiting places a cap on how many requests a client or user can send in a given time, which protects the system from sudden bursts and helps maintain availability for legitimate users. Pairing that with anomaly detection gives visibility into unusual or bot-like traffic patterns, allowing automated responses such as throttling, challenging, or blocking suspicious activity before it overwhelms the system. This approach directly tackles both the volume of incoming requests and the nature of the traffic, keeping AI services operational under stress. The other options don’t address availability: increasing data leakage controls focuses on confidentiality, reducing system monitoring reduces detection and response capability, and limiting access only during peak times unnecessarily hinders legitimate users and doesn’t reduce the attack impact.