Data poisoning is a difficult-to-detect AI attack that manipulates training data, retrieval pipelines, or third-party components to compromise model behavior, requiring continuous testing and runtime monitoring for defense.
Need help?
Contact usData poisoning is a difficult-to-detect AI attack that manipulates training data, retrieval pipelines, or third-party components to compromise model behavior, requiring continuous testing and runtime monitoring for defense.