HEFLGuard: Backdoor Detection in Homomorphic Encryption-Based Federated Learning

Summary

HEFLGuard is a backdoor detection framework for homomorphic encryption-based federated learning, where defenses that assume plaintext updates do not apply. The server builds overlapping validation models from encrypted client groups, and clients compare global and validation model logits on benign samples to expose backdoor behavior. Across seven vision and text benchmarks, three backdoor types and IID and non-IID settings, it reduces ASR from near 100% to near the no-backdoor level with clean accuracy dropping by no more than 2.5%.

Related items

• HighCVE-2026-123818: LMDeploy is a toolkit for compressing, deploying, and serving large language models. From 0.12.3 until 0.15.0…Similar attack · NVD/CVE Database
• HighGHSA-vvr2-h2jp-838m: Code Graph RAG: MCP paginated read_file path traversal / arbitrary file readSimilar attack · GitHub Advisory Database
• MediumAttackers Crash Claude's 'Office Party' Flaw to Run Malicious Code on LinuxSimilar attack · Dark Reading
• MediumCVE-2026-124588: huggingface finetrainers code injection via Library argumentSimilar attack · NVD/CVE Database
• MediumTaiwan-Based Hacker Used BOSSAI AI Agents to Steal Taipei Emergency Dispatch RecordsSimilar attack · The Hacker News