A regulator leak helps us understand how censorship works on the Russian internet

• The center is based in a modest, two-story red brick building in a business center on the banks of the Moscow River not far from the heart of the Russian capital. This division of Roskomnadzor is mostly tasked with ensuring the proper use of radio frequencies. However, it has another important yet unacknowledged responsibility – every day, it monitors the Russian internet for any content that could pose a reputational threat to the Russian authorities. Staff at the center seek out content that criticizes Putin or speculates on his health (one journalistic investigation reported that Russia’s president often travels with an oncologist, ear, nose and throat specialists and critical care doctors). Other hot topics include criticism of the war, rumors of mobilization, protests in the regions and negativity toward officials.
• Roskomnadzor systematically monitors internet resources to find this information, but its accuracy is limited. The department has commissioned Russia’s leading technical university, MFTI, to develop Vepr, a system for spying on social networks and media sites using artificial intelligence. They hope that this will do more than automatically flag “points of information conflict,” or hot topics at the national or regional level. They also want to predict the spread of “information threats” and the emergence of protests; publish automatic denials; block content and send data to “authorized bodies.” According to iStories, Vepr should rival the sophistication of China’s Great Firewall and is planned to go into service by the end of 2024. This deadline will almost inevitably be delayed due to problems with attracting qualified talent to work on the project amid the current sanctions.
• In leaked correspondence, the GRFC contacted Yandex, Russia’s leading tech company. Yandex denies doing Roskomnadzor any favors. However, internal documents show that GRFC used Toloka, Yandex’s platform for developing neural networks where users can analyze huge quantities of data at a low cost. In addition, GRFC’s “clean internet” service uses the Yandex search API to analyze online content (in simple terms, an API enables developers to use ready-made tools from other developers). GRFC staff asked Yandex to increase the number of possible API requests per day, but the company refused.
• All “negative” content found by the GRFC team is passed on to the security services: the Prosecutor General, the presidential administration, the Interior Ministry, the FSB, the FSO and the National Guard, reported iStories’ Alesya Marikhovskaya.
• In addition, the GRFC was reportedly preparing lists of foreign agents for the Justice Ministry long before this status was widely utilized in the latter half of 2021. For example, information about The Bell was compiled two years before we were officially recognized as a foreign agent. A significant part of that dossier has yet to be “used.” In Russia, the status of a foreign agent is used to discredit journalists, public figures, media outlets and human rights organizations that do not toe the official line.

Why the world should care

Analyzing these leaked documents helps us to understand exactly how censorship affects the Russian internet. Until now, we had no concrete details about the work of GRFC. Theoretically, any Russian citizen who criticizes the authorities online sees his activities passed through the state mechanism responsible for censorship. From there, the information is sent to the security forces. In this sense, the Russian internet is increasingly resembling its Chinese counterpart.