Research map / Large language models / Language, safety and interpretability
iSeal: Encrypted Fingerprinting for Reliable LLM Ownership Verification
Zixun Xiong, Gaoyi Wu, Qingyang Yu, Mingyu Derek Ma, Lingfeng Yao, Miao Pan, Xiaojiang Du, Hao Wang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| freeze the encoder during fingerprint training (ablation) | iSeal · Fingerprint success after equal epochs (LLaMA2-7B) | FSR: 0% -> 100% | Improved |
| use learned key-seeded encoder instead of AES (ablation) | iSeal · Fingerprint success after equal epochs (Mistral-7B) | FSR: 2% -> 100% | Improved |
| add Reed-Solomon error correction (ablation) | iSeal · Response manipulation attacks | FSR | Improved |
| inject fingerprint with iSeal instead of IF/WLM, under single-pair unlearning attack | iSeal · Collusion unlearning attack | FSR | Improved |
| inject fingerprint with iSeal instead of WLM/IF | iSeal · SuperGLUE 0-shot (LLaMA2-7B) | accuracy: 49% -> 56% | Improved |
| inject fingerprint with iSeal | LLaMA2-13B · SuperGLUE 0-shot | accuracy: 60% -> 59% | No clear effect |
| guess keys (random, random-key encoder, one-logit-off key) | iSeal · Key-guessing robustness | FSR: 0% | No clear effect |
| iSeal vs IF training time | iSeal · LLaMA2-13B convergence | minutes: 5 vs 5 | No clear effect |
Later papers that compared with this method
1 later accepted paper reports iSeal as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Echoes within the Reasoning: Stealthy and Effective Watermarking via Chain of Thought (ICML 2026 · better on 2 of 2 shared results)
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
- Fingerprinting Inference Systems of Large Language Models (NeurIPS 2026 · similarity 0.86)
- SIF: Semantically In-Distribution Fingerprints for Large Vision-Language Models (CVPR 2026 · similarity 0.86)
- Scalable Fingerprinting of Large Language Models (NeurIPS 2025 · similarity 0.86)
- HuRef: HUman-REadable Fingerprint for Large Language Models (NeurIPS 2024 · similarity 0.86)
- Intersecting-Boundary-Sensitive Fingerprinting for Tampering Detection of DNN Models (ICML 2024 · similarity 0.84)
- EncryIP: A Practical Encryption-Based Framework for Model Intellectual Property Protection (AAAI 2024 · similarity 0.84)
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.
