Fairness
Organisations should test whether AI outputs or decisions systematically disadvantage particular groups. Assessments should reflect the intended context of use rather than rely solely on overall accuracy.
Responsible AI practice IWIB4AI session Needs review
TEDx Alcoi Salon – El Poder de la IA: Responsible AI (Spanish)
Integrating AI into innovation while protecting privacy, reducing unfair bias and making its use transparent and inclusive.
Draft – legal and case law references not yet verified
Summary
This use case concerns organisations adopting AI to support innovation while embedding fairness, privacy and transparency into everyday practice. Responsible integration requires assessing how AI affects different people, explaining its role and protecting personal data. Inclusive design and staff training help ensure that innovation benefits users without overlooking those at greater risk of exclusion.
The Problem
AI can accelerate innovation, but adoption without adequate safeguards can reproduce discrimination, expose personal data and obscure how decisions are made. The challenge is to translate broad ethical commitments into practical checks throughout selection, deployment and ongoing use.
Ethical Grounds
Privacy
AI adoption should minimise personal data collection and prevent unnecessary disclosure through prompts, outputs or supplier access. Privacy safeguards should be built into tool selection and workflow design.
Transparency
People should understand when AI is being used, what role it plays and its relevant limitations. Explanations should be accessible to the people affected, not limited to technical documentation.
Inclusion
AI-enabled innovation should account for different languages, abilities and levels of digital literacy. Involving diverse users in design and testing helps identify barriers that standard performance measures may miss.
Legal Compliance
| Framework | Provision | Relevance |
|---|---|---|
| GDPR | Art. 5(1)(a)–(c) | Where AI processes personal data, organisations must ensure lawfulness, fairness and transparency, specify legitimate purposes and limit data to what is necessary. |
| GDPR | Arts. 13–14 | Controllers must provide the required information about personal data processing, including relevant AI-related processing, subject to the applicable conditions and exceptions. |
| GDPR | Art. 25 | Controllers must implement data protection by design and by default when integrating AI into personal data processing. |
| GDPR | Art. 35 | A data protection impact assessment is required before AI-related processing that is likely to result in a high risk to individuals’ rights and freedoms. |
| EU AI Act | Art. 4 | AI providers and deployers must take measures to ensure sufficient AI literacy among staff and others operating systems on their behalf, taking account of their knowledge, the use context and affected people. |
Sources
Court & Regulator Decisions
FILCAMS CGIL et al. v Deliveroo Italia Needs review
The discriminatory rider-ranking system shows why adoption checks must examine whether apparently neutral metrics disadvantage people with different circumstances.
CK v Dun & Bradstreet Austria (opens in a new tab)
The decision makes transparency operational by requiring explanations of automated decision procedures and principles, rather than allowing trade-secret claims to defeat access.
IMY – Skellefteå municipality facial recognition in school (opens in a new tab)
The school attendance decision demonstrates why organisations must assess necessity, privacy impacts and power imbalances before deploying biometric AI.
Key Takeaways
- Assess privacy, fairness and exclusion risks before adopting an AI tool.
- Minimise personal data use and configure privacy safeguards by default.
- Explain AI’s role and limitations in accessible language.
- Train staff and involve diverse users in testing and ongoing review.
Related Use Cases
From the Session
TEDx Alcoi Salon – El Poder de la IA: Responsible AI (Spanish)