This talk highlights the engineering controls and tool development to safeguard data across AI's lifecycle. Addressing vulnerabilities from collection to deployment, it emphasizes the creation of bespoke security practices, advanced access controls, and sophisticated anomaly detection tools. The presentation delves into the engineering practices essential for embedding security by design in GenAI systems, advocating for a proactive approach to data protection. It calls for cross-disciplinary collaboration in developing robust security frameworks and tools, ensuring AI applications are both powerful, secure and trustworthy.