The increasing integration of artificial intelligence (AI) into various sectors is transforming operations but simultaneously widening the cyber-attack surface. This article analyses the emerging risks of AI from a cybersecurity perspective, highlighting that AI, while a powerful defence tool, is also being exploited by malicious actors for sophisticated cyberattacks. We detail the vulnerabilities inherent in AI systems, the new offensive tactics it enables (notably deepfakes), and their implications for critical infrastructures and national security. The article highlights the importance of data protection, model integrity and abuse prevention, with reference to the NIST AI RMF framework for trusted AI systems. Finally, it explores the mitigation strategies and governance frameworks needed for resilient cybersecurity in the AI era, while addressing future challenges.