ATLAS Frameworks extend threat modeling to adversarial threats against which type of systems?

• A. AI Systems
• B. Cloud Services
• C. Mobile Devices
• D. Industrial Control Systems

Answer: (A)

ATLAS frameworks are designed to extend threat modeling to adversarial threats that specifically target AI systems. This matters because AI/ML components introduce unique attack surfaces beyond traditional IT—things like poisoning training data, evasion with adversarial inputs, model extraction, and membership or data leakage from models. ATLAS helps you map attacker goals, capabilities, and attack paths across data, models, and deployment environments, so you can reason about how an attacker could degrade model performance, steal intellectual property, or expose sensitive information. The other domains exist, but ATLAS focuses on adversarial risks inherent to AI systems, making AI the correct focus.