Users can influence AI models through deletion request behavior
Modeling Deletion Requests in Machine Unlearning
Machine LearningCryptography and SecurityComputers and Society
Summary
When people ask AI systems to delete their personal data, how they make those requests can change how the AI learns or forgets. The authors identified two ways users might ask for deletion: working alone or together and adapting based on responses. They found that these different behaviors can lead to different effects on the AI’s learning, especially when many users are involved. Using methods from data importance measurement, the authors showed how deletion requests could strategically shape model behavior in real-world image recognition tasks.
What this means in practice
- •For data privacy teams: Design deletion request policies to better control AI model behavior after data removal.
- •For machine learning engineers: Implement deletion requester models that influence AI retraining outcomes on vision tasks.
Authors
Christian Cianfarani, Aloni Cohen
Abstract
Machine unlearning is seen as a promising approach to enable users to exercise the "right to erasure" in the context of AI models. We ask how users might influence the behavior of models when exercising this right. We define two types of behaviors that users might adopt when requesting the deletion of their data: adaptivity and collectivity. Drawing connections between the goals of users in this context and results in stochastic optimization, we demonstrate theoretical gaps between the potential effects of groups of users who do and do not display these behaviors. We then show how techniques from data valuation might be used to design deletion requesters that can significantly alter model behavior in realistic settings. In experiments on computer vision tasks, we demonstrate the differential effects of different models of user behavior and attempt to isolate the impacts of adaptivity and collectivity.