InfoResearchPeer-reviewed
FUBA: Backdoor Federated Learning via Federated Unlearning
- Published
- Record updated
Summary
Researchers present FUBA, a backdoor attack on federated learning models that exploits federated unlearning. Adversarial clients inject a backdoor during training, then one client submits an unlearning request to remove its contribution, which activates the backdoor in the model. The authors report that FUBA reliably backdoors the model after unlearning regardless of the unlearning method used, and that it remains stealthy against existing backdoor defenses.
Related items
- InfoAI-Defined Vehicles: Securing the Decisions That Drive Vehicle BehaviorSimilar attack · VicOne Blog
- MediumResearchers hijack Hugging Face account for popular Kokoro AI modelSimilar attack · BleepingComputer
- Medium'Sleeper Agent' Tactic Hides Payloads in Open-Weight AI ModelsSimilar attack · Dark Reading
- MediumTrained to betray: How post-training supply chain attacks hijack coding agentsSimilar attack · Microsoft Security Blog
- MediumRewriting less than 10 neurons in an LLM can turn it into a sleeper backdoorSimilar attack · CSO Online