China will use AI to detect tax crimes

China is developing a system that uses artificial intelligence and big data to detect financial fraud, including tricks unknown to most tax inspectors.

Researchers believe that traditional taxthe system is too fragmented and cannot process all available data efficiently enough. Therefore, for three years, Aisino, together with scientists from Harbin Institute of Technology and Beijing University of Post and Telecommunications, have been developing an automated solution.

According to the team, during this time about 300 thousand.state inspectors took part in training the system algorithms. Currently, the program is able to detect more than 95% of tax crimes, including those that people would not even notice.

The developed mechanism has been integrateddirectly into the software used by the State Tax Service of China. The administration has not yet revealed the details of the project, but it is known that the effectiveness of the system has been confirmed by several pilot programs in economic centers in the east of the country. Although the government has not yet approved the full-blown launch, researchers are talking about achieving positive testing results.

The new system can be connected to everyonegovernment databases, including information on property, goods, international trade and business registration. This allows the technology to automatically detect whether companies and individuals have provided false information in  declarations, and also find new methods of evasion previously unknown to the authorities.

In the process, the system self-learns, defining new schemes, and constantly updates its algorithms in order to take into account the latest legislative changes.

Researchers expect that artificial intelligence will reduce the influence of the human factor in the field of taxation, and the growth in revenue in the treasury will determine the country's economic growth.

Despite the development of new technologies,New ways to bypass them are also emerging. Recently, researchers discovered that to deceive most recognition algorithms, an ordinary printed mask with the face of another person is enough.

</p>