Trustworthy AI: Definition and Examples

The concept of Trustworthy AI rests on the idea that an artificial intelligence system must not only be technically performant but also trustworthy on ethical, social, and legal levels. It is a holistic approach that encompasses algorithm transparency, fairness of outcomes, technical robustness, and respect for privacy.

This framework was widely popularized by the European Commission's guidelines in 2019, which identify seven key requirements: human oversight, technical robustness and safety, privacy, transparency, diversity and non-discrimination, societal and environmental well-being, and accountability. These principles now form the foundation of the European AI Act and influence global regulations.

In prompt engineering, the notion of Trustworthy AI translates into designing prompts that encourage verifiable, unbiased, and transparent responses. This involves explicitly asking the AI to cite its sources, signal its uncertainties, avoid stereotypes, and respect the limits of its knowledge rather than fabricating information.

The stakes are fundamental: as AI integrates into critical domains such as healthcare, justice, and finance, the trust placed in these systems must rest on concrete and measurable guarantees, not merely technological promises. Trustworthy AI thus represents a paradigm shift from AI optimized solely for performance to AI optimized for reliability and accountability.