Add description for adversarial-attacks topic

Author: Tariolle

topics/adversarial-attacks/index.md

@@ -0,0 +1,7 @@
+---
+display_name: Adversarial attacks
+short_description: Adversarial attacks are techniques that craft intentionally perturbed inputs to mislead machine learning models into producing incorrect outputs.
+topic: adversarial-attacks
+wikipedia_url: https://en.wikipedia.org/wiki/Adversarial_machine_learning
+---
+Adversarial attacks are techniques that craft intentionally perturbed inputs to mislead machine learning models into producing incorrect outputs. They are central to research in AI robustness, security, and trustworthiness.