This modernisation is reflected in particular by a 40% increase in the budget for IT services, in 2019 it should be remembered that this budget amounted to around 199 million euros.
This increase in the IT services budget allows DGFIP to develop and test new types of tax audits using artificial intelligence and data mining.
First of all, what is artificial intelligence? Artificial intelligence, also called AI, is a set of techniques capable of reproducing certain human faculties such as logic, reasoning, etc.
Secondly, what is data mining? Data-mining is a kind of technique for analysing a large volume of information / data. This process makes it possible in particular to detect anomalies, as well as correlations within a very large amount of data.
Thanks to these new technologies (for which DGFIP also employed a certain number of data scientists, IT specialists, technicians, programmers, etc.), targeted tax audits could be implemented during the year 2019.
As indicated in the DGFIP report on its activity during the year 2019, the use of these technological tools made it possible to analyse the data held by DGFIP with those held by external organisations (resulting in particular from the automatic exchange of information – we would refer you to our following article on this: l’échange automatique d’information). This in-depth analysis made it possible to match up a great deal of data, and to observe certain inconsistencies or certain fraudulent behaviour.
Consequently, 22% of tax audits were scheduled using these technological tools in 2019. Thanks in particular to data mining, DGFIP was still able to collect 785 million euros.
Company A has been subject to a tax audit following an AAG (“acte anormal de gestion” – abnormal management action). The company has subsequently undergone a tax audit, this company is known to DGFIP, and its data is stored in the DGFIP files.
We then have a company B, with no apparent link with company A (different manager and shareholders) located in another geographical area (another region). This company, whose activity is different from company A, has the same telephone number as company A.
The software and programs of DGFIP will thus be able to link these 2 companies to each other thanks to these new technologies, and therefore, since company A has been subject to a tax audit, company B, having a link with the latter can also be monitored by DGFIP.
Furthermore, if a company C has no link with company A, but one of its managers is the same as in company B, DGFIP, having identified company B as having a link with company A, may also monitor company C in view of the fact that the latter has a “link” with company B (same manager).
The link between these companies can thus be made very quickly, which would not necessarily have been the case previously, especially if these companies have registered offices located in different departments or regions.
Thanks to these tools, DGFIP can monitor “so-called risk companies”, while carrying out fewer audits than before.
Obviously, tests were carried out beforehand, the Regional Directorate of Public Finance of Brittany having been chosen to experiment with AI. As can be seen on the official economie.gouv website (https://www.economie.gouv.fr/aife/lia-et-datascience-pour-meilleur-controle-des-depenses-letat#), this experimental pilot had been a success; indeed it should be noted that, with the use of these new tools, about “80% and more of anomalies had been detected, whereas with an audit (so-called classic audit) the figure was only 40%”.
The use of these new technical processes will therefore certainly be successful.
However, Artificial Intelligence and data-mining will not only be used to track down irregularities committed by taxpayers, but these tools should also help the State to control its spending.
Indeed, a decree was issued on 29 January 2019, creating an automated processing of predictive analysis relating to the control of State expenditure. This automated processing, which will bring together a large amount of data, should help public accountants to identify certain elements (e.g. payment at risk, expenses that may give rise to irregularities, etc.).
This automated processing should thereby help public accountants with their assignments, while participating in better budget management.
Obviously, these technological tools have also been developed to help companies in difficulty.
As mentioned in its 2019 activity report, DGFIP has set up a tool (based on AI) in order to target companies in difficulty. All kinds of difficulties can therefore be detected (as mentioned in this report, this could be a risk of insolvency proceedings, or a risk resulting from financing difficulties). By targeting companies that are in difficulty, or which could be so in the near future, the administrative authorities will be able to provide support to these companies, and may be able to save some of them.
Contact Maître Benjamin A. Kergueno, Attorney at Law today if you are dealing with issues related to real estate law in France and on the French Riviera.
Maître Benjamin A. Kergueno, LL.M will provide you with a full set of informations and with the adequate counsels for sorting it out.
For more information or to schedule an appointment with an experienced lawyer regarding real estate law in France, please contact us.
