NIST Identifies Main Types of Adversarial Machine Learning Threats

By

Learn about four types of potential AI attacks.

A new National Institute of Standards and Technology (NIST) publication identifies general types of cyberattacks – so-called “adversarial machine learning” threats – that can be used to attack or manipulate the behavior of AI/ML systems.

The four main types, according to the news statement, are:

  • Evasion attacks, which attempt to alter an input after the AI is deployed.
  • Poisoning attacks, which occur in the training phase through the introduction of corrupted data.
  • Privacy attacks, which attempt to gain and misuse sensitive information about the AI or the data on which it was trained.
  • Abuse attacks, which involve malicious insertion of incorrect information into a source.

The publication “is intended to help AI developers and users get a handle on the types of attacks they might expect along with approaches to mitigate them – with the understanding that there is no silver bullet.”
 
 
 

 
 
 

01/26/2024
AI , Artificial Inte... , attack , cybersecurity , machine learning , Security

Related content

comments powered by Disqus