OD is used to train artificial intelligence models,

[tanjimajuha20]

"Theory is dry, my friend, but the tree of life grows green," wrote Goethe in "Faust." Indeed, theory often diverges from practice. In the case of the amendments to the legislation on the processing of anonymized personal data adopted in the summer, the same may happen. What is wrong with these innovations is discussed by Irina Levova, Director of Strategic Projects at the Internet Research Institute (IRI), and Karen Kazaryan, Director General of the IRI

Dive into "gosozero"

First, a little theory. indonesia telegram Anonymized data (AD) is mathematically processed personal data (PD), which as a result of this processing may remain personal if it does allow identifying a person, or may cease to be so if it does not allow. At the same time, whether it is possible to identify a person in a dataset or not is a probability that can be calculated. If the probability is extremely low, such data is not considered personal.

forecast, conduct research, combat fraud, and improve service security. For these purposes, the use of anonymized rather than personal data is primarily due to the reduction of leak risks and the protection of citizens' rights.

In general, the development of the bill was aimed at ensuring the "free circulation of anonymized data" (clause 1.3 of the federal project "Regulatory regulation of the digital environment"), but something went wrong during the work. As a result, this summeracceptedthe long-suffering law is not about OD, as was originally planned, but about the transfer of data from any operators to some as yet non-existent “lake” of data (state information system) for the “needs of improving public administration,” whatever meanings are put into this phrase.

Companies must transfer "anonymized personal data" to "gosozero." At the same time, the deputies included a very funny legal construction in the law. If you analyze it, it becomes completely unclear what exactly can or cannot be transferred to "gosozero."

Let's sort out the terms

The current definition of PD is extremely broad - it is "any information related to a directly or indirectly defined or determinable individual (subject of personal data)" (clause 1 of Article 3 No. 152-FZ). That is, strictly speaking, PD is anything that is somehow related to some person and that Roskomnadzor can call personal data. For example, any postal address or a mobile phone number written on a fence can be classified as such data.