Hello, you are using an old browser that's unsafe and no longer supported. Please consider updating your browser to a newer version, or downloading a modern browser.
Training Camp • Cybersecurity Glossary
Model poisoning corrupts an AI model's training data or process to degrade accuracy or implant backdoors, a key machine learning integrity risk.
Model Poisoning Definition: Model poisoning corrupts an AI model's training data or process to degrade accuracy or implant backdoors, a key machine learning integrity risk.
Model poisoning is an adversarial machine learning attack in which an attacker corrupts a model's training data or learning process to degrade its performance or implant hidden behaviors. Data poisoning manipulates training samples, while backdoor poisoning embeds triggers that cause targeted misclassification when activated. It is a significant supply-chain and integrity risk for AI systems, especially those trained on external or crowdsourced data.
Turn knowledge into credentials with our instructor-led cybersecurity boot camps.
View All Courses →