Responsible AI
Also called trustworthy AI, ethical AI
Responsible AI is the practice of designing, buying and using artificial intelligence so that it is fair, transparent, safe, secure and accountable to people. It turns principles such as those of the OECD into policies, controls and reviews that cover an AI system from design to retirement.
How it works
A regulated firm usually sets a responsible AI policy approved by the board, then builds it into existing risk processes. Each AI use case is assessed for bias, explainability, data protection and security before launch. Committees approve higher-risk uses. Models are monitored after deployment, and customers can challenge automated decisions through human review.
The term has no single legal definition, but its principles now sit inside binding law. The EU AI Act, data protection law and consumer protection rules all draw on fairness and transparency. Supervisors judge firms on outcomes, so a lending model that treats groups unfairly creates conduct and legal risk whatever label the firm gives its programme.
The rules
First intergovernmental AI standard, setting principles on fairness, transparency, robustness and accountability; revised in May 2024.
OECD members and adherent governmentsVoluntary framework organised around govern, map, measure and manage functions to build trustworthy AI characteristics into practice.
Any organisation designing, developing or using AI, on a voluntary basisSets risk-based legal duties for AI, including prohibitions, high-risk requirements and transparency duties, giving legal force to many responsible AI principles.
Providers and deployers of AI systems placed on or used in the EU marketCommon questions
Is responsible AI a legal requirement?
Not as a single term. Its principles appear in binding law such as the EU AI Act, data protection law and anti-discrimination rules. Frameworks such as the OECD principles and the NIST AI RMF are voluntary.
How is responsible AI different from AI governance?
Responsible AI describes the outcomes a firm wants, such as fairness and transparency. AI governance is the structure that delivers them, including policies, committees, inventories, approvals and monitoring across the life of each system.
Who owns responsible AI in a bank?
Ownership is usually shared. The business owns each use case, risk and compliance functions set standards and challenge, and internal audit tests the controls. Many firms name a senior executive accountable for AI risk.
