Trustworthy AI — Seven Requirements for AI Systems

Definition

Trustworthy AI refers to AI systems that exhibit properties that make them worthy of human and institutional trust: they are lawful (compliant with applicable regulations), technically robust (reliable, accurate, and secure), and ethically sound (fair, transparent, accountable, and respectful of fundamental rights). The concept was systematically developed by the European Commission’s High-Level Expert Group on AI (HLEG AI), which in 2019 published the Ethics Guidelines for Trustworthy AI establishing seven key requirements.

The seven requirements for trustworthy AI are: human agency and oversight; technical robustness and safety; privacy and data governance; transparency; diversity, non-discrimination and fairness; societal and environmental wellbeing; and accountability. These requirements are not merely aspirational principles — they are now embedded in binding legal obligations through the EU AI Act, which translates most of them into enforceable requirements for high-risk AI systems.

ISO/IEC 42001 provides the certifiable management system framework that operationalizes trustworthy AI as a governance practice rather than a communications statement.

Why it matters operationally

Trustworthiness in AI matters because AI systems that are not trusted are not used — or are used despite mistrust, which is more dangerous. The adoption and value extraction from AI depends on stakeholders — users, operators, clients, regulators, and the public — having justified confidence that the systems behave as expected, within defined parameters, and in alignment with values that protect rather than harm.

The challenge for organizations is operationalizing trustworthiness: moving from the seven HLEG AI principles to implemented controls that can be independently verified. An AI governance policy that cites the principles does not demonstrate trustworthy AI. A certified management system under ISO/IEC 42001, combined with ethical conformity verification through the Ethical AI Mark, provides the independent attestation that makes trustworthiness credible to external parties.

Regulatory framework

Framework	Trustworthy AI requirements
EU HLEG AI Guidelines (2019)	The seven requirements for trustworthy AI: human oversight, technical robustness, privacy and data governance, transparency, diversity and non-discrimination, societal and environmental wellbeing, and accountability.
EU AI Act	Converts HLEG AI requirements into legal obligations for high-risk systems. Most of the seven requirements have direct correspondence with specific Regulation obligations.
ISO/IEC 42001	The certifiable management system that operationalizes trustworthy AI requirements. Certification provides independent conformity evidence.
Ethical AI Mark (Zertia)	The conformity mark that verifies the ethical requirements of trustworthy AI against four ISO standards: ISO TR 24368, ISO TR 24027, ISO TR 24028, ISO TS 6254.

How Zertia evaluates it

Zertia’s certification and audit services together cover the full trustworthy AI framework. ISO/IEC 42001 certification validates the governance infrastructure: management system, risk processes, oversight mechanisms, and continual improvement. The Ethical AI Mark validates ethical conformity at the system level against ISO TR 24368 (ethical requirements), ISO TR 24027 (bias and equity), ISO TR 24028 (transparency), and ISO/IEC TS 6254 (explainability) — the four international standards that directly address the HLEG AI’s ethical requirements.

[ISO 42001 Certification] · Ethical AI Mark