Roskomnadzor is exploring the possibility of using artificial intelligence technologies for automatic detection of prohibited content in games.
It is assumed that trainable neural networks will be able to independently identify signs of possible violations, generate digital tags, and transmit relevant information to regulatory authorities for further verification.
For example, we have a platform with constantly updated content. Conditional Minecraft. And it is impossible to monitor it constantly. For this, it is possible to use machine learning technology, which, based on this sample, a trainable sample, will form markers of prohibited content that is distributed there.
The agency believes that such a mechanism can increase the effectiveness of detecting illegal materials in gaming services. Currently, the initiative is at the stage of discussion and preliminary study of possible implementation options.
At the same time, RKN still found a person (or a real expert was invited) who understands the technical features of games, who noted that an attempt to use neural networks on top of the game code would be registered by built-in anti-cheat systems and blocked, so the regulator proposes embedding AI algorithms directly into the architecture of games.