Summary
In spring 2021, The EU-Commission put forward a proposal for an EU Artificial Intelligence Act (AIA). The AIA is to enter into force in 2023.
As regard to public authorities’ use of Big Data and Artificial Intelligence in order to detect and prevent fraud, corruption, money laundering and related areas of organized economic crimes, the AIA will have a similar impact on the legal requirements as the GDPR had in 2018.
The aim of AICompare is to address the built-in uncertainty in the AIA and clarify the content of the AIA before it enters into force. AICompare will thus enable public authorities to comply with the AIA and avoid fines.
SSR
It is our view that the project will have significant societal impact. Both the Danish and Dutch tax authorities are facing massive (re-)investments in the development of AI to identify different types of fraud and economical crime. These solutions must be developed in compliance with the AIA in order to avoid cost of re-development and fines (as recently imposed on the Dutch tax authorities for violating GDPR in connection with the use of artificial intelligence to identify fraud, see https://gdprhub.eu/index.php?title=AP_(The_Netherlands)_-_Tax_Administration_fined_for_discriminatory_and_unlawful_data_processing&mtc=today ). Here, the public authorities will be able to use our research results to ensure compliance. A more indirect societal impact is related to trust. The increased use of data as well as the use of artificial intelligence to identify fraudsters has already caused considerable public debate. As an example, the chairman of the Danish Data Ethics Council have publicly expressed his serious concern, see e.g. https://www.altinget.dk/embedsvaerk/artikel/johan-busse-ny-lov-giver-skat-indsigt-i-hvor-meget-du-satte-din-kaffemaskine-til-salg-for-paa-den-blaa-avis . Therefore, we consider it extremely important that the authorities at least comply with the AIA, if citizens' trust in the public authorities is to be maintained.