CryptoForge: Adversarial Co-Evolution for Automated Cryptographic Hash Function Design

CryptoForge demonstrates that adversarial co-evolution is a viable approach to automated hash function design. A neural attacker finds weaknesses; a spectral engine provides cheap algebraic estimates; an evolutionary defender modifies the design; the cycle repeats. Over 8 generations, the system independently converged on the AES S-box. The binary security cliff at NL ≈ 80 and the R² collapse are empirical findings about the structure of the design space that, to our knowledge, have not been previously reported. The source code, trained models, web interface, and all 17 result files are publicly available.¹

[1] X. Wang, Y. L. Yin, and H. Yu, “Finding collisions in the full SHA-1,” in Advances in Cryptology—CRYPTO 2005, LNCS 3621, Springer, 2005, pp. 17–36. [2] A. Gohr, “Improving attacks on round-reduced Speck32/64 using deep learning,” in Advances in Cryptology—CRYPTO 2019, LNCS 11693, Springer, 2019, pp. 324–352. [3] A. Benamira, D. Gerault, T. Peyrin, and Q. Q. Tan, “A deeper look at machine learning-based cryptanalysis,” in Advances in Cryptology—EUROCRYPT 2021, LNCS 12696, Springer, 2021, pp. 805–835. [4] S. Picek, K. Knezevic, D. Jakobovic, and J. F. Miller, “Evolving S-boxes with improved cryptographic properties,” in Proc. Genetic and Evolutionary Computation Conf. (GECCO), ACM, 2016, pp. 409–410. [5] C. Carlet, Boolean Functions for Cryptography and Coding Theory. Cambridge University Press, 2010. [6] J. Daemen and V. Rijmen, The Design of Rijndael: AES—The Advanced Encryption Standard. Springer, 2002. [7] C. Lu, C. Lu, R. T. Lange, J. Foerster, J. Clune, and D. Ha, “The AI Scientist: Towards fully automated open-ended scientific discovery,” arXiv:2408.06292, 2024. [8] A. Rukhin et al., “A statistical test suite for random and pseudorandom number generators for cryptographic applications,” NIST Special Publication 800-22, Rev. 1a, 2010. [9] M. Matsui, “Linear cryptanalysis method for DES cipher,” in Advances in Cryptology—EUROCRYPT ’93, LNCS 765, Springer, 1993, pp. 386–397. [10] E. Biham and A. Shamir, “Differential cryptanalysis of DES-like cryptosystems,” J. Cryptology, vol. 4, no. 1, pp. 3–72, 1991.