AI SECURITY • 360 / 397
Understand how untrusted text, tools and autonomy create new attack paths.
Adversarial Example
An adversarial example is an input crafted to cause a model to make an incorrect prediction while appearing normal or minimally changed.
Think of it like
Think of Adversarial Example as an input-trust or privilege problem: words can influence software decisions, so boundaries must be explicit.
Real life
Email, documents and web pages can become hostile inputs to AI systems through attacks involving Adversarial Example.
SRE lens
Security classifiers and vision systems can be tested against adversarial inputs.
Remember thisAn adversarial example is an input crafted to cause a model to make an incorrect prediction while appearing normal or minimally changed.