User Isolation Poisoning on Decentralized Federated Learning: An Adversarial Message-Passing Graph Neural Network Approach

Summary

Researchers propose user isolation poisoning (UIP), an attack on decentralized federated learning in which a malicious user distributes compromised model updates while following the standard protocol. The attack uses an adversarial message-passing graph (MPG) neural network, implemented in PyTorch, to degrade benign users' embeddings and their influence on the shared model. The authors report a 49.5% drop in test accuracy and say the attack evades existing cosine similarity- and Euclidean distance-based defenses.

Related items

• MediumDetecting sleeper agents hiding in AI model neuronsSimilar attack · Microsoft Security Blog
• CriticalCVE-2026-48705: MONAI PydicomReader affine computation accepts non-finite DICOM valuesSimilar attack · NVD/CVE Database
• InfoSigned but Steerable: Measuring and Closing the Text-Vector Integrity Gap in Offline Retrieval-Augmented GenerationSimilar attack · Arxiv (cs.CR + cs.CL + cs.LG)
• InfoSecuring next-generation edge intelligence: A survey from federated learning to autonomous agents in cyber-physical systemsSimilar attack · OpenAlex (peer-reviewed AI security)
• InfoPop Goes the Verifier: Phoneme-Targeted Data Poisoning of Voice Liveness DetectionSimilar attack · Arxiv (cs.RO + cs.CV security)