Theoretical foundations of trust in digital evidence in the context of the development of artificial intelligence technologies

The article examines the problem of ensuring trust in digital evidence in the context of the rapid development of artificial intelligence technologies and generative neural network models. The subject of the research is the theoretical foundations of such trust, as well as the fundamental limitations of modern expert methods for verifying electronic files used in criminal and civil proceedings. The methodological basis of the research is an interdisciplinary approach that integrates methods of system analysis, digital forensics and comparative legal research. A structural analysis of four groups of digital file verification methods has been performed: EXIF metadata analysis, file structure research, system log verification, and device identification. Specific vulnerabilities have been identified for each group that arise when using generative adversarial networks (GANS), specialized metadata manipulation tools, and methods to circumvent system limitations of mobile devices.

The scientific novelty of the work lies in the formalization of a model of multi-stage falsification of a digital object, which includes three successive stages: generative image synthesis with imitation of the characteristics of a specific device, deep manipulation of EXIF metadata with complete overwriting of time, geolocation and hardware attributes, as well as file embedding on the target device using privileged access and synchronization of system timestamps. For the first time, a comprehensive analysis of the technological asymmetry between the means of falsification and verification of digital evidence has been conducted, demonstrating that each examination method can be countered by an available countermeasure that neutralizes its effectiveness.

Based on the results obtained, the necessity of introducing into judicial practice the principle of the presumption of the dubiousness of digital evidence, which implies a mandatory comprehensive study of electronic files indicating the limits of applicability of expert methods. Practical recommendations are formulated to improve the regulatory framework, develop multidisciplinary fraud detection programs, and modernize expert standards for assessing the authenticity of electronic evidence.

1. Герасимчук М. Ю., Герасимчук Е. В. Недееспособность и ограниченная дееспособность лиц в рамках наследственных правоотношений // Наследственное право. – 2022. – № 1. – С. 12–15. – URL: https://www.elibrary.ru/item.asp?id=48015787

2. Карцхия А. А., Макаренко Г. И. Правовые проблемы применения искусственного интеллекта в России // Правовая информатика. – 2024. – № 1. – С. 4–19.

3. Кондрашов П. Е., Петрунин Ю. Ю., Попова С. С. Регулирование разработки и применения искусственного интеллекта: проблемы и направления развития // Вестник Московского университета. Серия 21. Управление (государство и общество). – 2025. – № 1. – С. 3–19.

4. Кустова М. Н., Умарова А. А. Этика и инновации: исследование этических аспектов влияния инновационных технологий на современное общество // Региональная и отраслевая экономика. – 2025. – № 4. – С. 48–57.

5. Подузова Е. Б. Права на аккаунт при использовании технологий искусственного интеллекта: нотариальные и цивилистические аспекты защиты // Нотариальный вестник. – 2023. – № 6. – С. 30–37.

6. Репин Д. А. Технологии искусственного интеллекта как фактор совершенствования государственного управления: вызовы и угрозы. Экономика и управление. – 2025. – № 32(2). – С. 139–148.

7. Третьякова О. Ю. Применение технологий искусственного интеллекта в нотариальной деятельности: проблемы и перспективы // Нотариальный вестник. – 2023. – № 12. – С. 58–64.

8. Deng, P. Exploration of Scientific and Technological Ethics Governance in the Age of Artificial Intelligence, Journal of Electronic Research and Application. – 2025. – Vol. 9. Is. 4. – P. 97–103.