Practical Secrets Extraction against Black-box LLMs

arXiv:2609.36941 · cs.CR · Submitted 2026-09-29 · Read on arXiv

cs.CR

Submitted: 2026-09-29

Updated: 2026-09-29

Code: https://github.com/zai-org/feedback

Terminology

• Distillation — 8× in this paper · explained in 2 episodes
• Alignment — 5× in this paper · explained in 1 episode
• Calibration — 5× in this paper · explained in 7 episodes
• Hallucination — 5× in this paper · explained in 4 episodes
• Knowledge Distillation — 5× in this paper · explained in 11 episodes
• Real Rate — 5× in this paper
• Benchmark — 4× in this paper · explained in 2 episodes
• Fine-Tuning — 3× in this paper · explained in 5 episodes
• Ablation — 2× in this paper · explained in 1 episode
• Consistency — 2× in this paper · explained in 3 episodes
• IEEE Symposium on Security and Privacy — 2× in this paper
• Large Language Models (LLMs) — 2× in this paper · explained in 6 episodes
• Structural Constraints — 2× in this paper · explained in 2 episodes
• Two-Stage Framework — 2× in this paper · explained in 2 episodes
• Verification — 2× in this paper · explained in 2 episodes
• Ablation study — 1× in this paper · explained in 1 episode
• Actor — 1× in this paper · explained in 3 episodes
• Adam — 1× in this paper · explained in 2 episodes
• Adaptation — 1× in this paper · explained in 1 episode
• Computational overhead — 1× in this paper · explained in 2 episodes
• Core — 1× in this paper · explained in 2 episodes
• Cosine Similarity — 1× in this paper · explained in 3 episodes
• Decode — 1× in this paper · explained in 2 episodes
• Grounding — 1× in this paper · explained in 3 episodes
• In Network and Distributed System Security Symposium — 1× in this paper
• Recovered Secrets — 1× in this paper
• Trace — 1× in this paper · explained in 2 episodes

Sources

• The Llama 3 Herd of Models
• DeepSeek-Coder: When the Large Language Model Meets Programming -- The Rise of Code Intelligence
• Qwen2.5-Coder Technical Report
• Training Data Leakage Analysis in Language Models
• StarCoder 2 and The Stack v2: The Next Generation
• Understanding Privacy Risks in Code Models Through Training Dynamics: A Causal Approach
• Seed-Coder: Let the Code Model Curate Data for Itself

Related papers

• SoK: AI-Augmented Binary Reversing
• Relaxed Sender Anonymity for CBDC Interbank Settlement: A Zero-Knowledge Approach on Permissioned EVM
• Calibration-Family Overfit: Why Trusted Sabotage Monitors Don't Transfer Across Lineages
• Efficient Fuzzy PSI under One-Sided Assumptions
• Sealing the Audit-Runtime Gap for LLM Skills
• Token Composition: A Graph Based on EVM Logs