Unlearning Incentivizes Learning under Privacy Risk

While machine learning empowers intelligent services and offers users customized experiences, privacy concerns emerge from regulatory requirements and the privacy-conscious demands of users. Machine unlearning presents a potential solution to these concerns. Despite the growing demand for practical deployment due to the right to be forgotten privacy regulations, the economic impact of machine unlearning on user behavior and platform profitability remains largely unexplored and may limit its implementation. In this paper, we formulate a set of contract design problems under both unlearning-disabled and unlearning-enabled scenarios. Challenges arise when the unlearning-enabled platform jointly designs compensation for both learning and unlearning to incentivize users' sequential decisions to balance the expected revenue and unlearning cost. We first conduct a questionnaire survey that reveals that machine unlearning increases users' willingness to participate in federated learning. We then provide a necessary condition for maximizing the surplus of an unlearning-enabled platform, enabling the point-wise decomposition for the optimal contract design problem, based on which we minimize the incentive cost and maximize the surplus for the platform. Our further analysis reveals that i) the incentive effects of unlearning grow quadratically with users' privacy sensitivity, and ii) enabling unlearning may even profit more than disabling it, under higher cost elasticity of risk distribution. Our numerical results show that the platform's profitability is primarily influenced by users' privacy sensitivity. When users are relatively highly privacy-sensitive, enabling unlearning can significantly improve profitability. CCS Concepts • Security and privacy → Economics of security and privacy; • Computing methodologies → Model development and analysis.