Glossary
Adversarial Machine Learning
Adversarial ML is tricking AI by giving it confusing inputs - like putting a sticker on a stop sign that makes self-driving cars ignore it.
What is Adversarial Machine Learning?
Techniques to manipulate or deceive machine learning models through specially crafted inputs.
Why Should You Care?
ML systems in security can be evaded. Attackers craft inputs that bypass ML-based detection systems.
Is your business exposed?
Real-World Example
Malware authors modify code slightly to evade ML-based antivirus while maintaining functionality.
How to Protect Against Adversarial Machine Learning
- 1.
Test ML security models against adversarial samples
- 2.
Implement adversarial training for ML models
Related Terms
Is your business exposed?
Check if your company data is circulating on the dark web
Free scan • No credit card required