Fairness Principle

UK AI regulatory principle requiring that AI systems not undermine legal rights, discriminate unfairly against individuals, or create unfair market outcomes. Actors across the AI lifecycle must consider definitions of fairness appropriate to the system's use, outcomes, and applicable law.

Legal Framework Integration:

• Equality and human rights law (Equality Act 2010, Human Rights Act 1998)
• Data protection requirements (UK GDPR, Data Protection Act 2018)
• Consumer and competition law protections
• Public Sector Equality Duty for public authorities
• Sector-specific fairness requirements (e.g., FCA Handbook)

Implementation Requirements:

• Regulators develop sector-specific descriptions of fairness
• Joint guidance where multiple regulatory domains intersect
• Consideration of relevant technical standards (ISO/IEC TR 24027:2021, ISO/IEC 12791, ISO/IEC TR 24368:2022)
• Justifiable decisions for high-impact AI-enabled outcomes

Context-Specific Application: Fairness definitions must be appropriate to specific sectors and use cases. For example, AI recruitment tools require different fairness considerations than financial credit scoring systems. Regulators must collaborate where AI applications span multiple domains.

Rationale: AI can significantly impact people's lives through insurance offers, credit scores, and recruitment outcomes. High-impact AI decisions should not be arbitrary and must be justifiable. The principle ensures compliance with existing equality, data protection, and consumer protection laws while addressing AI-specific fairness challenges.

Regulators must interpret fairness within their domains while ensuring consistency across the regulatory landscape through collaboration and joint guidance where appropriate.