Detecting financial fraud and tax evasion is complex and requires data from a variety of sources. This may include bank deposits and financial transactions, investments in shares, investments in real estates, expenditure on cell phones, import and export information, car insurances, etc., and an understanding of the tax system. The complexity is realised when the relationship between various features of the data is not always explicit, consequently it is difficult for humans to manually identify tax fraud, tax evasion and the loopholes in the tax system. The use of artificial intelligence to analyse this "big data" has the potential to transform traditional ways of identifying financial fraud and tax evasion. This talk demonstrates the complexities which most tax systems face, and discusses the role of artificial intelligence in reducing financial fraud and tax evasion.