Our work on attribute unlearning for user-level federated recommender systems has been accepted to AAAI 2026.