Keys in the Weights: Transformer Authentication Using Model-Bound Latent Representations
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866911450518257664 |
|---|---|
| author | Okatan, Ayşe S. Akbaş, Mustafa İlhan Kandel, Laxima Niure Peköz, Berker |
| author_facet | Okatan, Ayşe S. Akbaş, Mustafa İlhan Kandel, Laxima Niure Peköz, Berker |
| contents | We introduce Model-Bound Latent Exchange (MoBLE), a decoder-binding property in Transformer autoencoders formalized as Zero-Shot Decoder Non-Transferability (ZSDN). In identity tasks using iso-architectural models trained on identical data but differing in seeds, self-decoding achieves more than 0.91 exact match and 0.98 token accuracy, while zero-shot cross-decoding collapses to chance without exact matches. This separation arises without injected secrets or adversarial training, and is corroborated by weight-space distances and attention-divergence diagnostics. We interpret ZSDN as model binding, a latent-based authentication and access-control mechanism, even when the architecture and training recipe are public: encoder's hidden state representation deterministically reveals the plaintext, yet only the correctly keyed decoder reproduces it in zero-shot. We formally define ZSDN, a decoder-binding advantage metric, and outline deployment considerations for secure artificial intelligence (AI) pipelines. Finally, we discuss learnability risks (e.g., adapter alignment) and outline mitigations. MoBLE offers a lightweight, accelerator-friendly approach to secure AI deployment in safety-critical domains, including aviation and cyber-physical systems. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2511_00973 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Keys in the Weights: Transformer Authentication Using Model-Bound Latent Representations Okatan, Ayşe S. Akbaş, Mustafa İlhan Kandel, Laxima Niure Peköz, Berker Cryptography and Security Artificial Intelligence Signal Processing 68T07, 68P25, 94A60, 68Q32, 68Q87, 94A17, 68M12 I.2.6; E.3; D.4.6; C.2.0; I.5.1; C.2.2; K.6.5; C.3 We introduce Model-Bound Latent Exchange (MoBLE), a decoder-binding property in Transformer autoencoders formalized as Zero-Shot Decoder Non-Transferability (ZSDN). In identity tasks using iso-architectural models trained on identical data but differing in seeds, self-decoding achieves more than 0.91 exact match and 0.98 token accuracy, while zero-shot cross-decoding collapses to chance without exact matches. This separation arises without injected secrets or adversarial training, and is corroborated by weight-space distances and attention-divergence diagnostics. We interpret ZSDN as model binding, a latent-based authentication and access-control mechanism, even when the architecture and training recipe are public: encoder's hidden state representation deterministically reveals the plaintext, yet only the correctly keyed decoder reproduces it in zero-shot. We formally define ZSDN, a decoder-binding advantage metric, and outline deployment considerations for secure artificial intelligence (AI) pipelines. Finally, we discuss learnability risks (e.g., adapter alignment) and outline mitigations. MoBLE offers a lightweight, accelerator-friendly approach to secure AI deployment in safety-critical domains, including aviation and cyber-physical systems. |
| title | Keys in the Weights: Transformer Authentication Using Model-Bound Latent Representations |
| topic | Cryptography and Security Artificial Intelligence Signal Processing 68T07, 68P25, 94A60, 68Q32, 68Q87, 94A17, 68M12 I.2.6; E.3; D.4.6; C.2.0; I.5.1; C.2.2; K.6.5; C.3 |
| url | https://arxiv.org/abs/2511.00973 |