Login Get a Demo
01 Glossary · Governance and risk

What Is Responsible AI?

The principles behind the phrase, three public texts to start from, and how responsible AI differs from trustworthy AI, governance and compliance.

The short answer

What is responsible AI?

Last updated
Read time
4 min read
From
Cranium AI
Also called
RAI
Share on LinkedIn

Responsible AI is the practice of designing, building, buying and using AI systems in ways that are lawful, safe, fair, transparent and accountable to the people they affect.

It is a goal stated in principles. AI governance and AI risk management are the mechanisms that turn those principles into decisions and evidence.

Key takeaways
  • NIST uses the term without a formal definition, tying it to human centricity, social responsibility and sustainability.
  • The OECD AI Principles, adopted in 2019 and updated in 2024, have 47 adherents and five values-based principles.
  • NIST lists seven characteristics of trustworthy AI, from valid and reliable to fair with harmful bias managed.
  • A principle counts only when it has an owner, a control and a record.
By the numbers Each figure links to its sourceAll sources 
  • 47

    adherents to the OECD AI Principles, adopted in 2019 and updated in 2024

    Source: OECD.AI

  • 7

    characteristics of trustworthy AI set out in NIST AI RMF 1.0

    Source: NIST AI 100‑1

  • 7

    key requirements in the EU Ethics Guidelines for Trustworthy AI, presented April 8, 2019

    Source: European Commission

02 The principles

Which principles define responsible AI?

Three public texts are the usual starting points. None of them is law.

01

OECD AI Principles

Five values-based principles: inclusive growth, sustainable development and well-being; human rights and democratic values, including fairness and privacy; transparency and explainability; robustness, security and safety; and accountability.

OECD AI Principles
02

NIST AI RMF

Seven characteristics of trustworthy AI: valid and reliable; safe; secure and resilient; accountable and transparent; explainable and interpretable; privacy-enhanced; and fair with harmful bias managed.

NIST AI RMF
03

EU Ethics Guidelines

The 2019 expert group guidelines call trustworthy AI lawful, ethical and robust, and set seven requirements that open with human agency and oversight and close with accountability.

Ethics Guidelines
03 Compare

Responsible AI vs. trustworthy AI vs. AI governance

TermWhat it isWhat it produces
Responsible AIThe goal: AI that is lawful, safe, fair, transparent and accountablePrinciples and commitments
Trustworthy AIThe properties a system should have, such as NIST’s seven characteristicsTest results and evaluations for each property
AI governanceThe decision rights, policies and controlsPolicies, an inventory, approvals and reviews
AI risk managementThe process that finds, scores and treats riskA risk register and treatment plans
AI complianceMeeting a specific law or standardEvidence mapped to articles and clauses
04 In practice

How do you put responsible AI into practice?

  1. Adopt a principle set and an owner

    Start from a published set, not a blank page, and make one executive answerable for it.

  2. Turn each principle into a requirement

    Fairness becomes a bias test before release. Transparency becomes documentation a user can read. Accountability becomes a named owner and traceable decisions, as OECD Principle 1.5 describes.

  3. Check it at intake and assessment

    Apply the requirements when a use is proposed and when it is assessed, so a principle is met before launch, not audited after it.

  4. Set objectives and report

    ISO/IEC 42001 includes controls on objectives for the responsible development and responsible use of AI systems (Annex B.6.1.2 and B.9.3). Report against them.

05 Keep going

Related reading.

06 Questions

Frequently asked questions.

Straight answers, kept current. Last updated .

Ask us anything →
Is responsible AI a legal requirement?

Not by that name. The NIST AI RMF is voluntary and the EU’s 2019 ethics guidelines are guidance. The EU AI Act turns several of the same ideas into binding duties for high-risk AI, including a risk management system (Article 9) and human oversight (Article 14).

What is the difference between responsible AI and trustworthy AI?

Responsible AI describes how an organization behaves. Trustworthy AI describes what a system is like. NIST uses the two together, writing that trustworthy AI systems and their responsible use can mitigate negative risks. In practice the terms travel as a pair, and the three texts above cover the same ground under either name.

What is the difference between responsible AI and AI governance?

Responsible AI is the aim and AI governance is the machinery. Principles say AI should be fair and accountable. Governance assigns the owner, sets the policy, keeps the inventory and requires the assessment that shows whether a given system is. One without the other is either a statement nobody can verify or a process with no purpose.

Does NIST define responsible AI?

Not formally. The AI RMF says responsible AI practices can help align decisions about AI system design, development and use with intended aims and values, and that the core concepts in responsible AI “emphasize human centricity, social responsibility, and sustainability”. It then defines trustworthiness in detail, through seven characteristics.

07 See it for yourself

From the definition
to the evidence.

See how Cranium AI discovers, observes, governs, secures and proves every AI model, agent and vendor in your enterprise.