Which principle emphasizes integrating security considerations from the start of AI development rather than as an afterthought?

• A. Data quality in AI security
• B. Human-in-the-loop in AI
• C. Secure by design in AI architecture
• D. Chairperson

Answer: (C)

Secure by design in AI architecture means making security a fundamental design constraint from the very beginning of development. This approach requires considering potential threats, privacy, data handling, access controls, and resilience during the planning, design, and implementation stages so security features and safeguards are built into the system rather than added later. By modeling threats, planning defensive measures, and validating security throughout the development lifecycle, the AI system becomes more robust against attacks such as data poisoning, model leakage, and prompt manipulation, while also supporting governance and compliance goals.

The other ideas don’t capture that same proactive integration. Data quality in AI security focuses on ensuring the input data is clean and trustworthy rather than embedding security into the architecture from the start. Human-in-the-loop emphasizes human oversight and control, which is important but not the same as designing security into the system from inception. The chairperson option is unrelated to the technical principle of building security into the AI design.