Task Statement 5.1: Explain methods to secure AI systems.

Securing AI systems requires a multi-layered approach encompassing data integrity, privacy, infrastructure protection, and regulatory compliance. This involves implementing secure data engineering practices like encryption, access control, and data quality checks, along with privacy-enhancing technologies such as differential privacy and anonymization. Application security must be reinforced with input validation, API protection, and vulnerability management. AI-specific threats like prompt injection in generative models demand input sanitization and guardrails. Tools like Amazon Macie, KMS, GuardDuty, and SageMaker Clarify help automate detection, enforce security, and maintain trust across the AI lifecycle. Ensuring transparency, source citation, and documentation further strengthens accountability and reduces legal risk.