Cyber-physical systems rely enormously on secure communication particularly as they progress to incorporate current cutting-edge capabilities in artificial intelligence. These systems, increasingly used in critical areas such as the healthcare sector, manufacturing, and transportation, feature physical processes mixed with computational intelligence and networked communication. Since CPSs rely fundamentally on real-time data transmission as well as decision-making that is reliant upon such information for its accuracy, potential security breaches will have catastrophic impacts. This chapter discusses many related issues and fixes toward safe communication for AI-powered CPS. Important security issues include risks to data availability, confidentiality, and integrity as well as issues with authorisation and authentication. To show how AI can be both an enabler and a potential risk factor for CPS security, vulnerabilities brought about by its integration, such as adversarial attacks, model inversion, and data poisoning, are examined. This chapter covers the basics of secure communication, including cryptographic techniques, secure key management plans, and protocols like TLS and IPsec, which are designed for resource-constrained and real-time settings. It also demonstrates how AI is revolutionising the security of CPS communications. While adaptive security mechanisms utilise real-time threat intelligence to automatically change communication paths, AI-based intrusion detection systems (IDS) use machine learning to detect unusual network activity.