Which of the following BEST ensures the integrity of datasets used to train AI models?
Cryptographic integrity controls, such as hashes and digital signatures, allow datasets to be tracked and verified so unauthorized modification can be detected. NIST notes that providers can publish cryptographic hashes and downloaders can verify training data; it also defines integrity protection as assurance that data has not been altered without authorization.
Community Discussion