Adversarial ML

Adversarial ML. Testing AI models by attempting to trick them with adversarial inputs to bypass security filters or extract training data.