The Federation Council discussed the use of AI and digitalization tools to combat corruption

On Tuesday, a meeting was held at the Federation Council site on the use of artificial intelligence (AI) and digitalization tools in the field of anti-corruption. As it turned out from the speeches of representatives of the Russian Academy of National Economy and Public Administration (RANEPA), the Prosecutor General’s Office and the Accounts Chamber, anti-corruption digitalization is already producing significant results, but it is advisable to reinforce this success with certain legislative changes.

The meeting was opened by Oksana Shmaliy, Deputy Director of the Institute of Law and National Security of RANEPA, who spoke about the work of the state information system (GIS) in the field of anti-corruption “Poseidon”. Let us remind you that the system was created by decree of President Vladimir Putin last spring. With its help, you can collect and analyze information about the income and property of officials.

According to Ms. Shmaliy, the Poseidon GIS is focused on the prevention of corruption offenses and uses data from the Federal Tax Service, the Federal Property Management Agency and Financial Monitoring, as well as information from social networks. “Its use allows you to create a kind of digital profile of the person being checked and obtain information about possible conflicts of interest: informal communication with representatives of controlled citizens, organizations, and so on,” she explained.

However, the mere fact of Poseidon’s discovery of possible corruption correlations is not a basis for bringing the person being inspected to justice – only a “signal” for a possible, but optional verification, Ms. Shmaliy emphasized. In practice, “some problems” may arise here, which requires legislative support for a special anti-corruption check in such cases, the expert added.

Next, Oksana Shmaliy touched upon developments related to the use of AI in the work of the polygraph and control and supervisory activities, and also complained about the lack of the necessary legal framework and general legislation on artificial intelligence: “It could resolve questions about the subjectivity of the AI ​​system, the admissibility of using the results its activities as evidence, guarantees of the rights of citizens in respect of whom verification activities are carried out using AI, and so on.” In her opinion, the “Digital Code” and the federal law “On Digital Profile” could correct the situation, and a temporary “palliative” could be the introduction of appropriate amendments to the anti-corruption legislation.

The head of the intellectual analytics laboratory of the Center for Digital Solutions and AI at RANEPA, Sergei Bolovtsov, spoke about the development of a system for analyzing draft regulatory legal acts for corruption factors – selective changes in the scope of law, excessive freedom of subordinate rulemaking, a document going beyond its competence, incompleteness of administrative procedures, refusal of competitive procedures and even legal and linguistic uncertainty. According to him, the assessment of the reliability of the examination carried out by this system already reaches 75%.

In turn, Aslan Yusufov, Deputy Head of the Department for Supervision of the Execution of Anti-Corruption Legislation of the Prosecutor General’s Office, spoke about the German experience in using AI to identify cases of money laundering. The efficiency of such a system reaches 95%. Mr. Yusufov also touched upon the topic of cryptocurrencies in criminal activity: according to him, contrary to popular belief, only 0.29% of their turnover is used for criminal purposes.

Finally, the chief inspector of the Accounts Chamber, Olga Samarina, spoke about the use of AI to analyze the plans of federal executive authorities to combat corruption. A study of more than 100 documents showed that some of them do not correspond to national plans in this area, and labor costs for this work were reduced by 15–20 times. If inspectors would spend 300–400 man-days analyzing documents manually, then with the help of AI, all the work would be completed in 20, Ms. Samarina noted.

Andrey Prah