AAIR: Advanced in AI Risk

By Isaca · Question Mode

QuestionQ42

AI Life Cycle Risk Management

An organization relies on multiple external data sources to train its AI models. Which of the following is the risk practitioner’s BEST recommendation for protecting the organization against data-poisoning attacks?

Explanation

Data poisoning occurs when an adversary controls or inserts malicious training data. Continuous monitoring and anomaly detection in data-ingestion pipelines provide the most direct control by detecting abnormal or potentially poisoned inputs before they are incorporated into training data and influence the model.

Learn more

Community Discussion

No comments yet. Be the first to start the discussion!