Abilkair A.K.
ARTIFICIAL INTELLIGENCE AS A DEFENSIVE TOOL AGAINST CYBERATTACKS: MACHINE LEARNING-BASED KEY RECOVERY ANALYSIS OF THE ELGAMAL CRYPTOSYSTEM *
Аннотация:
as artificial intelligence technologies continue to evolve, their application in the field of cybersecurity opens both new opportunities and risks. This study examines the potential of machine learning algorithms to recover private keys in the ElGamal cryptosystem—a widely used method in public-key encryption. Using synthetic datasets generated from ElGamal key pairs, we applied various supervised learning models to predict private keys based on public values. The results show that neural networks achieved the highest prediction accuracy in simplified scenarios, while the success rate dropped significantly with longer key lengths. These findings highlight the dual role of AI in enhancing and potentially challenging cryptographic security. The research concludes that, although machine learning can model mathematical relationships in encryption under limited conditions, strong cryptographic parameters such as increased key size remain effective in protecting against AI-based attacks.
Ключевые слова:
cybersecurity, artificial intelligence, machine learning, ElGamal cryptosystem, secret key prediction, cryptanalysis
ACKS: MACHINE LEARNING-BASED KEY RECOVERY ANALYSIS OF THE ELGAMAL CRYPTOSYSTEM WWАннотация: as artificial intelligence technologies continue to evolve, their application in the field of cybersecurity opens both new opportunities and risks. This study examines the potential of machine learning algorithms to recover private keys in the ElGamal cryptosystem—a widely used method in public-key encryption. Using synthetic datasets generated from ElGamal key pairs, we applied various supervised learning models to predict private keys based on public values. The results show that neural networks achieved the highest prediction accuracy in simplified scenarios, while the success rate dropped significantly with longer key lengths. These findings highlight the dual role of AI in enhancing and potentially challenging cryptographic security. The research concludes that, although machine learning can model mathematical relationships in encryption under limited conditions, strong cryptographic parameters such as increased key size remain effective in protecting against AI-based attacks.Ключевые слова: cybersecurity, artificial intelligence, machine learning, ElGamal cryptosystem, secret key prediction, cryptanalysis.TTIntroduction.In the rapidly evolving landscape of cybersecurity, the race between attackers and defenders is increasingly shaped by advancements in artificial intelligence (AI). With the rise of more complex and adaptive threats, traditional encryption techniques are now being scrutinized under new analytical lenses. One such method, the ElGamal cryptosystem, has long been considered a reliable asymmetric encryption standard due to its dependence on the hardness of the discrete logarithm problem. However, the emergence of AI-based cryptanalysis introduces novel challenges to even the most established cryptographic methods.Artificial intelligence, particularly through machine learning (ML), has already demonstrated its transformative impact across industries—from medical diagnostics to financial forecasting. In the domain of cybersecurity, ML is primarily used to enhance threat detection and response mechanisms. But beyond defense, its capabilities raise important questions about potential offensive uses. One such scenario involves key recovery attacks, where ML models attempt to predict or reconstruct private cryptographic keys based solely on public information—a task once considered infeasible without brute-force techniques.This study investigates the feasibility of applying supervised machine learning algorithms to recover private keys in the ElGamal encryption scheme. By generating datasets composed of public-private key pairs, we train multiple ML models—namely, Linear Regression, Decision Trees, and Random Forests—to approximate the private key from its public counterpart. These experiments simulate an adversarial scenario where an attacker, without direct access to the secret key, leverages statistical patterns to compromise encrypted communications.Figure 1. AI-driven key recovery using ML models to predict private keys from public inputs in the ElGamal cryptosystem. As shown in Figure 1, the input public key is fed into various ML models—Linear Regression, Decision Tree, and Random Forest—which attempt to estimate the corresponding private key. The primary aim of this study is to assess the feasibility of such recovery under constrained experimental conditions and to evaluate the impact of factors such as key length and dataset size on model performance. This approach aims to contribute to the ongoing discourse about the robustness of classical encryption in the face of emerging intelligent threats.Materials and methods.Dataset Description.To evaluate the performance of machine learning algorithms in recovering private keys in the ElGamal cryptosystem, we generated synthetic datasets simulating the key generation process. Each sample was constructed as a tuple (y, x), where
Номер журнала Вестник науки №6 (87) том 3
Ссылка для цитирования:
Abilkair A.K. ARTIFICIAL INTELLIGENCE AS A DEFENSIVE TOOL AGAINST CYBERATTACKS: MACHINE LEARNING-BASED KEY RECOVERY ANALYSIS OF THE ELGAMAL CRYPTOSYSTEM // Вестник науки №6 (87) том 3. С. 1616 - 1622. 2025 г. ISSN 2712-8849 // Электронный ресурс: https://www.вестник-науки.рф/article/24408 (дата обращения: 17.02.2026 г.)
Вестник науки © 2025. 16+