Ghost in the Transformer: Detecting Model Reuse with Invariant Spectral Signatures
Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (40) · 33648-33656
DOI: 10.1609/aaai.v40i40.40654 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Suqing Wang und 3 weiteren Autor:innen.
Abstract (Originalsprache)
Large Language Models (LLMs) are widely adopted, but their high training cost leads many developers to fine-tune existing open-source models. While most adhere to open-source licenses, some falsely claim original training despite clear derivation from public models, raising pressing concerns about intellectual property protection and the need to verify model provenance. In this paper, we propose GhostSpec, a lightweight yet effective method for verifying LLM lineage without access to training data or modification of model behavior. Our approach constructs compact and robust fingerprints by applying singular value decomposition (SVD) to invariant products of internal attention weight matrices. Unlike watermarking or output-based methods, GhostSpec is fully data-free, non-invasive, and computationally efficient. Extensive experiments show it is robust to fine-tuning, pruning, expansion, and adversarial transformations, reliably tracing lineage with minimal overhead. By offering a practical solution for model verification, our method contributes to intellectual property protection and fosters a transparent, trustworthy LLM ecosystem.
Zitieren
APA
Wang, S., Ma, Z., Xinyi, L., & Li, Z. (2026). Ghost in the Transformer: Detecting Model Reuse with Invariant Spectral Signatures. Proceedings of the AAAI Conference on Artificial Intelligence, 40(40), 33648-33656. https://doi.org/10.1609/aaai.v40i40.40654
Vancouver
Wang S, Ma Z, Xinyi L, Li Z. Ghost in the Transformer: Detecting Model Reuse with Invariant Spectral Signatures. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(40):33648-33656. doi:10.1609/aaai.v40i40.40654
BibTeX
@article{Wang2026,
author = {Wang, Suqing and Ma, Ziyang and Xinyi, Li and Li, Zuchao},
title = {Ghost in the Transformer: Detecting Model Reuse with Invariant Spectral Signatures},
journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
year = {2026},
volume = {40},
number = {40},
pages = {33648--33656},
doi = {10.1609/aaai.v40i40.40654}
}
EUVIMED