ZJU's PGPO obfuscates embeddings to defend LLM recommender inversion attacks
zju · hf · 2026-11-19
Zhejiang University proposes PGPO (Prototype-Guided Progressive Obfuscation) to defend LLM-enhanced recommender systems against embedding inversion attacks. It identifies high-influence prototype words, uses GRPO-guided LLM variant exploration and anchor-oriented graph expansion, with rewards balancing embedding dissimilarity, structural consistency, and semantic constraints — perturbing high-influence semantic embeddings while preserving relative item relations.
More from Safety
- Human Principal launches private beta: cryptographic proof a human is behind your AI agent — yenkel · 2026-11-19
- US AI lawsuits hit 336 cases, with OpenAI facing 119 of them — emmanuelvivier · 2026-11-19
- Citi makes customers liable for third-party AI agents and account hacks — emmanuelvivier · 2026-11-19
- AI safety debate: sprint on 2027 takeoff fears or stretch the timeline to 2040 — sayashk · 2026-11-19
- UK MPs move to criminalise misuse of AI — HotMinuteUK · 2026-11-19
- ZK Podcast: ARIA's Nicola Greco on generative cryptography and AI scaling trust — DavideCrapis · 2026-11-19