The Greek tax authority AADE is being equipped with three new digital “scanners” that will detect potential violations in real-time, utilizing artificial intelligence (AI), digital data, and advanced analytical techniques. The “smart” tools were presented at a special AADE event, attended by authority officials and representatives from the European Union and the Swedish Tax Authority. These include:
1.AI predictive models: A pilot AI and machine learning system will prioritize tax audit cases based on the probability of tax deviation. This way, audits will focus where risk is highest, utilizing modern prediction techniques integrated into risk analysis.
2.Next-generation risk analysis: This “super weapon” enables audit mechanisms to focus on high tax risk cases. The system introduces unified case selection criteria and combines data from Taxis, myDATA, Property Registry, banks, GEMI and EFKA. Additionally, random sampling maintains the element of unpredictability, enhancing the accuracy and effectiveness of tax audits.
3. E-commerce and cryptocurrency guide: The new manual standardizes audit steps for digital platforms and cryptocurrency transactions, emphasizing undeclared income, VAT discrepancies, and digital traces of illegal transactions. This gives AADE tools to monitor e-commerce and the digital economy in real-time.
Combating tax evasion: How AADE’s “scanners” are fed and operate
The new technological tools give AADE the ability to utilize data that was previously lost in the maze of traditional audits. Continuous data flows from myDATA, Digital Customer Registry, Digital Shipping Document, and Electronic Invoicing allow audit services to observe the market and transactions in real-time. These systems now also function as “alarm bells” for targeted audits. Operationally, case selection now occurs through a centralized system that “scores” risks for cases under review, based on multiple criteria analysis, providing complete oversight and supervision of audits to AADE management. The system is based on a combination of models for targeted audits by sector, depending on available reliable data, while maintaining traditional random sampling to identify new forms of tax evasion outside the program.
