The AOC Consortium (Open Administration of Catalonia) has developed an innovative solution using generative artificial intelligence to facilitate and streamline the anonymization of personal data in municipal council minutes. This tool aims to reduce manual workload and minimize the risk of errors in publishing sensitive information.
The solution stands out for its ability to interpret document context, going beyond simple pattern detection. This allows it to distinguish between data that should be public, such as the name of an elected official, and information requiring anonymization, like personal details of other individuals present in the minutes. The AI automatically suggests what information to protect, always under final user supervision.
The process is intuitive: users upload the minutes, define anonymization criteria, and the system generates a proposal. Subsequently, users review, modify if necessary, and download the document ready for publication. This human supervision model combines AI efficiency with the legal judgment of public administrations.
The proof of concept, validated with 52 documents from nine town councils, has shown an overall accuracy of 96% in identifying data to be anonymized. The results confirm the technical viability of the proposal and its potential to optimize personal data management in local government entities.
In parallel, La Meva Salut is transitioning to idCAT Mòbil as its primary identification system. This measure aims to simplify and enhance the security of accessing personal health information.
