# Learning to Protect Communications with Adversarial Neural Cryptography

arXiv: https://arxiv.org/abs/1610.06918 · HTML: https://mlsanity.com/paper/1610.06918/

## Key facts

- Cited by 11 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 21 Oct 2016.

## Built on by (most cited first, 11 of 11)

- [Hidden in Plain Text: Emergence & Mitigation of Steganographic Collusion in LLMs](https://mlsanity.com/paper/2410.03768/) (2024)
- [The Threat of Offensive AI to Organizations](https://mlsanity.com/paper/2106.15764/) (2021)
- [Noiseless Privacy-Preserving Decentralized Learning](https://mlsanity.com/paper/2404.09536/) (2024)
- [Odysseus: Jailbreaking Commercial Multimodal LLM-integrated Systems via Dual Steganography](https://mlsanity.com/paper/2512.20168/) (2025)
- [Seed-Induced Uniqueness in Transformer Models: Subspace Alignment Governs Subliminal Transfer](https://mlsanity.com/paper/2511.01023/) (2025)
- [Mind the (optimality) Gap: A Gap-Aware Learning Rate Scheduler for Adversarial Nets](https://mlsanity.com/paper/2302.00089/) (2023)
- [Keys in the Weights: Transformer Authentication Using Model-Bound Latent Representations](https://mlsanity.com/paper/2511.00973/) (2025)
- [Channel model for end-to-end learning of communications systems: A survey](https://mlsanity.com/paper/2204.03944/) (2022)
- [Neural Networks Meet Elliptic Curve Cryptography: A Novel Approach to Secure Communication](https://mlsanity.com/paper/2407.08831/) (2024)
- [Privacy-Preserving Federated Learning via Homomorphic Adversarial Networks](https://mlsanity.com/paper/2412.01650/) (2024)
- [SpinML: Customized Synthetic Data Generation for Private Training of Specialized ML Models](https://mlsanity.com/paper/2503.03160/) (2025)

## Abstract (arXiv)

We ask whether neural networks can learn to use secret keys to protect information from other neural networks. Specifically, we focus on ensuring confidentiality properties in a multiagent system, and we specify those properties in terms of an adversary. Thus, a system may consist of neural networks named Alice and Bob, and we aim to limit what a third neural network named Eve learns from eavesdropping on the communication between Alice and Bob. We do not prescribe specific cryptographic algorithms to these neural networks; instead, we train end-to-end, adversarially. We demonstrate that the neural networks can learn how to perform forms of encryption and decryption, and also how to apply these operations selectively in order to meet confidentiality goals.