Research map / Large language models / Reasoning and chain of thought
SLIM: Secure and Efficient Inference for Large Language Models on Untrusted Devices via TEEs
Wei Wang, Zihao Guan, Xing Zhou, Yan Ding, Yusong Tan, Jie Yu, Bao Li
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes SLIM, a scheme for secure and efficient LLM inference on untrusted devices that uses TEEs together with weight obfuscation to prevent weight extraction.
Problem
Deploying LLMs on untrusted hardware risks weight extraction and unauthorized replication; existing TEE-based weight obfuscation schemes struggle to provide both strong security guarantees and high performance.
Approach
Uses Trusted Execution Environments with obfuscated model weights to protect LLM weights during inference on untrusted devices, aiming to combine strong security with low overhead (details not available in the provided abstract).
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use SLIM instead of per-vector scaling/permutation obfuscation (SOTER, ShadowNet, TransLinkGuard, CoreGuard) | TEE obfuscation schemes · GPT2-Small MNLI after statistical attack | attack-recovered accuracy (%): 80.94 -> 40.65 (CoreGuard vs SLIM) | Improved |
| increase Householder block size k beyond 32 (ablation) | SLIM · Qwen3 GoEmotions, LLaMA3.2 FinQA | attack-recovered accuracy | Hurt |
| use block sizes 2, 8 or 32 (ablation) | SLIM · Qwen3 GoEmotions, LLaMA3.2 FinQA | defense effect | No clear effect |
| increase number of transformation rounds (ablation) | SLIM · Qwen3 GoEmotions, block size 32 and 64 | model security | Improved |
| increase block size k (throughput vs KL divergence) (ablation) | SLIM · Throughput with 128-token inputs | throughput / KL divergence | Mixed |
| propagate authorization through layers (one TEE interaction) instead of per-layer TEE interaction | SLIM vs ArrowCloak/GroupCover · TTFT, four causal LMs | TTFT: 4.49x-13.80x (ArrowCloak), 5.68x-21.31x (GroupCover) overhead relative to SLIM | Improved |
| use SLIM with orthogonal mixing in FP16 | SLIM vs ArrowCloak/GroupCover · Top-1 token accuracy, FP16 | Top-1 token accuracy: ArrowCloak 0% -> SLIM 95.2% on Gemma3 | Improved |
| increase model size from GPT2-Small to Qwen3-4B | SLIM · TTFT ratio to Pure-GPU | TTFT ratio: 3.66x -> 1.38x | Improved |
| give attacker up to 60% of data for Full-Tuning or LoRA fine-tuning (ablation) | SLIM · Qwen3 GoEmotions, LLaMA3.2 FinQA, GPT2 MNLI, Gemma3 WIC | attack-recovered accuracy | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- CoreGuard — CoreGuard: Safeguarding Foundational Capabilities of LLMs Against Model Stealing in Edge Deployment (NeurIPS 2025) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- BiShield-TEE: On the (In-)Security of Unilateral Weight Obfuscation in On-Device TEE-Shielded LLM Partition (NeurIPS 2026 · similarity 0.89)
- Hidden No More: Attacking and Defending Private Third-Party LLM Inference (ICML 2025 · similarity 0.89)
- TensorCommitments: A Lightweight Verifiable Inference for Large Language Models (NeurIPS 2026 · similarity 0.88)
- Split-and-Denoise: Protect large language model inference with local differential privacy (ICML 2024 · similarity 0.87)
- TOPLOC: A Locality Sensitive Hashing Scheme for Trustless Verifiable Inference (ICML 2025 · similarity 0.87)
- Jailbreak Antidote: Runtime Safety-Utility Balance via Sparse Representation Adjustment in Large Language Models (ICLR 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
