| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 771.86 KB | Adobe PDF |
Orientador(es)
Resumo(s)
Ao longo da última década, o impacto da era BigData fez-se sentir no setor público Europeu e as Administrações Fiscais não são exceção. Uma das tendências mais relevantes consiste no uso de algoritmos de gestão de risco para estabelecimento de perfis de contribuintes. Este tipo de tratamento de dados, porém, não está isento de riscos, e os casos holandeses SyRI e Toeslagenaffaire puseram a nu as falhas do movimento de governança fiscal algorítmica. Inclusive, recentemente, a Comissão Europeia, na Proposta de Regulamento Inteligência Artificial, classificou os algoritmos de gestão de risco para estabelecimento de perfis como de alto risco para os direitos e garantias do titular de dados (in casu, contribuintes). Partindo de uma premissa de equilíbrio, i.e., de compreensão da premência do aprofundamento do processo de digitalização tributária e da necessidade de salvaguarda dos direitos e garantias dos contribuintes, nesta dissertação analisamos o regime previsto no Regulamento Geral de Proteção de Dados para as decisões automatizadas e, bem assim, a sua adequação para regular esta tipologia de tratamento dos dados dos contribuintes. Assim, percorremos brevemente os conceitos de Inteligência Artificial, BigData e Machine Learning, estabilizando-os para efeito deste estudo. De seguida, destacamos o Princípio da Transparência, pela sua relevância na tutela dos direitos dos administrados/contribuintes, o qual nos permite recentrar a investigação nas normas do Regulamento Geral de Proteção de Dados relevantes para o tema, por se apresentar como único instrumento na União Europeia e no ordenamento nacional especialmente vocacionado para a regulação desta matéria. Por fim, apontamos soluções em três planos diferentes (operacional, algorítmico e regulatório) para um futuro de governança fiscal ponderado.
Over the last decade, the impact of the BigData era has been felt in the European public sector, and Tax Administrations are no exception. One of the most relevant trends consists of the use of risk management algorithms for taxpayer profiling. This type of data processing, however, is not without risks, and the Dutch cases SyRI and Toeslagenaffaire have laid bare the flaws of the algorithmic tax governance movement. Even more so, in the proposed Artificial Intelligence Regulation, the European Commission has classified risk management algorithms for profiling as high risk to the rights and guarantees of the data subject (in casu, taxpayers). On a balanced basis, i.e., understanding the urgency of intensifying the tax digitalisation process, and at the same time the need to safeguard the rights and guarantees of taxpayers, in this dissertation we analyse the regime provided for in the General Data Protection Regulation for automated decisions and its conformity to regulate this type of data processing. Therefore, after the introduction, we briefly study the concepts of Artificial Intelligence, BigData and Machine Learning, soothing them for the exclusive purpose of this study. We then address the Principle of Transparency, for its importance in protecting the rights of the taxpayers, allowing to refocus the investigation on the relevant rules of the General Data Protection Regulation to the theme, since it is the only instrument in the European Union and national law specially designed to regulate this matter. Lastly, we recommend solutions in three different levels (operational, algorithmic and regulatory) for a future of well-adjusted algorithmic tax governance.
Over the last decade, the impact of the BigData era has been felt in the European public sector, and Tax Administrations are no exception. One of the most relevant trends consists of the use of risk management algorithms for taxpayer profiling. This type of data processing, however, is not without risks, and the Dutch cases SyRI and Toeslagenaffaire have laid bare the flaws of the algorithmic tax governance movement. Even more so, in the proposed Artificial Intelligence Regulation, the European Commission has classified risk management algorithms for profiling as high risk to the rights and guarantees of the data subject (in casu, taxpayers). On a balanced basis, i.e., understanding the urgency of intensifying the tax digitalisation process, and at the same time the need to safeguard the rights and guarantees of taxpayers, in this dissertation we analyse the regime provided for in the General Data Protection Regulation for automated decisions and its conformity to regulate this type of data processing. Therefore, after the introduction, we briefly study the concepts of Artificial Intelligence, BigData and Machine Learning, soothing them for the exclusive purpose of this study. We then address the Principle of Transparency, for its importance in protecting the rights of the taxpayers, allowing to refocus the investigation on the relevant rules of the General Data Protection Regulation to the theme, since it is the only instrument in the European Union and national law specially designed to regulate this matter. Lastly, we recommend solutions in three different levels (operational, algorithmic and regulatory) for a future of well-adjusted algorithmic tax governance.
Descrição
Palavras-chave
Governança fiscal algorítmica Regulamento geral da proteção de dados Inteligência artificial Bigdata Machine learning Transparência Estabelecimento de perfis Garantias dos contribuintes Opacidade Direito à explicação Algorithmic tax governance General data protection regulation Artificial intelligence Transparency Profiling Taxpayer guarantees Opacity Right to explanation