07/12/2025
The AEPD’s Policy on the Use of Artificial Intelligence: an introductory summary
By Concepción Campos
The Spanish Data Protection Agency has published its General Policy on the Use of Generative Artificial Intelligence in its administrative processes, a document designed to analyse how a public institution can incorporate AI in a responsible and ethical manner, whilst ensuring the protection of fundamental rights.
Far from presenting AI as a threat, the AEPD takes a clear stance: AI is already a structural tool in the functioning of public administrations, and its integration must be carried out in accordance with criteria of security, traceability and governance.
What this policy aims to achieve
The document centres on six strategic objectives:
- To improve the efficiency and quality of public services.
- To guarantee the protection of personal data and fundamental rights.
- To safeguard sensitive and confidential information.
- To protect the working conditions of public sector staff.
- Ensuring the continuity of administrative processes.
- Strengthen public confidence in the Administration.
According to this policy, innovation only makes sense if it is trustworthy, explainable and under human control.
In which processes will generative AI be applied?
The AEPD identifies numerous real-world use cases in the conduct of administrative activities, all of which are aimed at supporting the civil service rather than replacing human judgement. These include, in particular:
- Summarising and structuring public documents.
- Machine translation of texts.
- Drafting of reports, communications and decisions.
- Legal and regulatory assistance.
- Case file management and classification of complaints.
- Support with data protection impact assessments.
- Production of training and informational content.
A key principle is that there will be no automated decisions with legal implications without prior human oversight.
Risk analysis: AI also has its limits
The policy identifies and classifies the risks associated with the use of AI in administrative processes, including, amongst others: risks to fundamental rights; algorithmic bias and discrimination; leaks of confidential information; technological dependence on suppliers; lack of system explainability; security breaches or loss of consistency in administrative action. For the purposes of managing these risks, it stipulates that each specific use of AI must be accompanied by an individualised risk assessment.
Clear governance: allocation of responsibilities and rules of use for staff
The policy sets out a governance framework, treating AI as just another critical system and establishing strict rules for staff, stipulating that AEPD staff must:
- Use only authorised systems.
- Not enter personal or confidential data into external tools.
- Always cross-check and validate results generated by AI.
- Report if content has been generated using AI systems.
- Undergo compulsory ongoing training.
Furthermore, every system must be registered, documented and monitored on a regular basis.
Conclusion: AI is not a technological fad, but a 21st-century public infrastructure, and we must ensure its ethical and responsible use by adopting tools that provide an organisational and operational framework suited to each organisation, as the AEPD has done. View the policy via the link below.
Source: https://concepcioncampos.org/
