This article presents an analysis of the current challenges in detecting attacks on Artificial Intelligence (AI) systems. The use of AI has revolutionized information processing and analysis, but it also increase the chance of security breaches and privacy concerns. Adverse attacks aim to exploit vulnerabilities in AI systems, and their sophistication is constantly increasing. The complexity of AI systems and their lack of transparency make it difficult to detect attacks, and existing defense mechanisms may not be sufficient. Additionally, collecting tagged data to detect attacks can be difficult, and distinguishing attacks from normal behavior can also pose a challenge. Thus, this article provides a comprehensive analysis of the problems and offers solutions to address the challenges.
Сивков Д.И. (науч. рук. Воробьева А.А.) Problems related to the detection of attacks on artificial intelligence (AI) systems // Сборник тезисов докладов конгресса молодых ученых. Электронное издание. – СПб: Университет ИТМО, [2023]. URL: https://kmu.itmo.ru/digests/article/9794