In ML backdoor attacks, triggers can be what?

• A. Explicit Patterns Only
• B. Subtle Triggers Only
• C. Triggers Can Be Explicit Or Subtle
• D. Triggers Have No Impact On Behavior

Answer: (C)

In ML backdoor attacks, triggers are patterns or inputs that cause the model to behave in a targeted way when they appear. These triggers can be explicit, visible patterns like a sticker or patch on an image, or they can be subtle, barely perceptible perturbations or rare phrases/tokens in data. Attackers can design either type, and during training the model learns to associate the trigger with the attacker-chosen output. That’s why the best answer is that triggers can be explicit or subtle. The other options miss the reality that backdoor triggers come in a spectrum and do influence the model when present.