Abstract
This paper explores the integration of Artificial Intelligence (AI) into cybersecurity, highlighting its role in detecting, preventing, and responding to increasingly sophisticated cyber threats. A review of the state of the art demonstrates significant progress in anomaly detection, explainable AI (XAI), and resilience against adversarial attacks. The study examines practical applications of AI across multiple domains, including threat detection, automated incident response, vulnerability analysis, anti-phishing systems, and adaptive authentication. Commercial platforms such as Darktrace, Vectra AI, and Microsoft Defender showcase the real-world impact of AI-driven security solutions. Additionally, relevant research initiatives, such as the DARPA Cyber Grand Challenge and EU’s AI4Cyber project, emphasize the growing importance of AI in protecting critical infrastructures. A case study on detecting and containing an Advanced persistent Threat (APT) in the financial sector further illustrates the effectiveness of unsupervised learning and automated response mechanisms. Despite these advances, challenges remain, including false positives, lack of transparency, adversarial vulnerabilities, and ethical considerations. The paper concludes by identifying future directions, such as the development of robust explainable models, integration of generative AI, and closer collaboration between human experts and AI systems.