Protecting AI Models from Data Poisoning
Data Protection News
In clean-label attacks, attackers modify the data in ways that are difficult to detect. ReversingLabs reported an increase in threats—more than 1300%—circulating through open source repositories from 2020 to 2023.3 Data injection introduces fabricated data points to the training dataset, often to steer the AI model’s behavior in a specific direction. But because the AI...