The III DPD Network Meeting served to present the progress of the AOC's algorithmic transparency report (TRAL), sharing its approach with data protection professionals. The Working Group on Ethical and Responsible AI has been developing this report for a year, aiming to create a reference model that combines understandable information for citizens with verifiable accountability regarding the risks, guarantees, and governance of artificial intelligence (AI) systems.
Organized by the APDCAT, the meeting brought together Data Protection Officers (DPOs) to address challenges such as digital regulation and AI. Carme Callés, Head of Innovation Service, and Patricia Mallén, DPD at AOC, led a practical workshop and a presentation on how to make the risks and guarantees of AI systems visible, understandable, and verifiable.
During the participatory workshop 'From Risk to Guarantee,' fifteen participants analyzed a practical case involving the use of generative AI to anonymize municipal council minutes. Risks identified included the failure to detect personal data, the reversal of anonymization, over-reliance on AI, and unauthorized reuse of documents. Proposed measures included human review, blocking publications without validation, user training, and contractual guarantees.
In the afternoon, the new TRAL model was presented, structured in two layers: one for citizens, explaining AI use clearly, and another for governance and accountability, detailing risks, operation, and supervision. This model complements other assessments like DPIAs or FRIAs.
The meeting facilitated the comparison of the model with professionals and continued the development of a collective, multidisciplinary tool adaptable to public administrations, aiming to transform AI governance into effective and meaningful transparency for citizens.
