Swati Vijay Pulate

The increasing integration of Artificial Intelligence (AI) into critical systems has revolutionized industries, enabling unprecedented advancements in automation, decision-making, and efficiency. However, the very reliance on AI introduces unique vulnerabilities, making these systems attractive targets for cyber threats. From adversarial attacks designed to manipulate machine learning models to data poisoning and model inversion attacks, the security challenges facing AI systems are multifaceted. This paper explores the emerging landscape of cyber threats to AI systems and investigates robust security strategies to mitigate these risks. Leveraging advanced cryptographic techniques, anomaly detection algorithms, and adversarial training, this research highlights how to safeguard AI's integrity, confidentiality, and availability. By addressing both proactive and reactive defense mechanisms, the findings provide a comprehensive roadmap for securing AI systems against evolving cyber threats.. Keywords: AI security, cyber threats, adversarial attacks, data poisoning, model integrity, AI vulnerability, secure machine learning, adversarial training, anomaly detection, cryptographic techniques.

https://doi.org/10.58357/jedcs.2025.v4.i1.18