Russians warned about "indirect" fakes created by AI

Russians have begun to encounter a new kind of "indirect" fakes, which are created by neural networks trained on a foreign cultural code. Yulia Ablets, Deputy General Director of ANO Dialog Regions, founder of the Workshop of New Media, told about this on June 8 during the St. Petersburg International Forum (SPIEF).

"We are starting to face a new kind of fake, let's call them indirect fakes," she told RIA Novosti.

According to Ablets, modern fakes are no longer limited to deliberate distortion of facts. Neural networks create texts, videos, and images where the facts may be correct, but the context, values, and interpretation of events are shaped by the culture in which the model was trained.

It clarifies that the neural network "draws in" the values and norms of behavior of the culture in which it was trained, and then automatically transfers them into the product being created. This leads to the fact that the text, although it looks reliable from the point of view of facts — with the correct dates and geography — contains hidden attitudes that can distort the perception of events.

Such fakes, according to Ablets, encourage people to make wrong decisions.

Neural network business: the AI market in Russia is growing faster than the rest of the IT sector

However, domestic developers are still lagging behind Western ones — why is this happening

Mikhail Salkin, a lawyer in civil and criminal matters, said on March 25 that the owners of content generation platforms using artificial intelligence (AI) are not responsible for the harm that can be caused by created images or videos. Cybersecurity expert Alexander Dvoryansky recalled that attackers actively use videos generated by neural networks. The quality of such videos evolves very quickly, he noted.