Over the next three years, the Russian government will spend 30 billion rubles on the development of artificial intelligence (AI). And that’s just the publicly available data. There are classified budget items that provide for similar expenditures regarding military plans and the development of capabilities for the war against Ukraine. These figures were cited by Mr. Volodymyr Solovian, head of the Hybrid Warfare Analytical Group at the Ukraine Crisis Media Center, during an expert discussion entitled “Russian Generative AI as a Tool for Information Influence.”
The Kremlin began articulating this topic at the national discourse level in 2019 with the creation of the National AI Development Program. The following year, the federal Artificial Intelligence project was launched, and it was during this period that significant funds began to be allocated from the state budget for the development of this sector. Education became one of the focus areas. Leading universities are receiving funding, as the plan is to train at least 10,000 specialists in this field by 2030.
As of 2023, the Russian Federation has launched and tested the widespread adoption of large Russian AI language models (GigaChat, Yandex GPT, and the voice assistant Alisa).
“The regime is considering the possibility of creating an information bubble, within which it plans to establish total control over its own population. The latest legislative act in this area is the draft Federal Law on Supporting the Development of Technology and Artificial Intelligence in the Russian Federation,” said Volodymyr Solovian.
Ms. Daria Opryshko, an expert with the NGO Human Rights Platform and a member of UNESCO’s “Information for All” program, urged caution regarding any Russian innovations, even though Russia lags behind the U.S. by a factor of 660 in terms of computing power.
“They indeed passed this bill, but what’s most interesting is that even in their own discussions in Russia, they emphasize that they may potentially face problems in implementing it. And it’s not just about the technological gap. The requirements imposed on the datasets used to train the models will significantly limit the possibilities. So, we’ll have to wait and see how it turns out for them,” she added.
Establishing control over the domestic audience is just one of the tasks facing Russian developers. Russia is already using AI as a tool for disinformation at the operational level. Mr. Igor Samokhodsky, founder of Policy Genome and a researcher on the impact of generative AI, explained how this works.
“You can ask the same question in Russian, Ukrainian, or English. In English, you’ll get no response, but in Russian, you’ll get direct disinformation. And this isn’t just about the AI itself, but about the layer built on top of it. Even when Russian AI knows the correct answer, there’s a layer that filters responses through propaganda,” the expert explained.
According to various studies, between 29% and 37% of people today use large language models instead of Google for searches. Mr. Andriy Olenin, an analyst at Mantis Analytics and head of the Eastern Europe division, pointed out the importance of what is known as fake balance in this context.
“We can ask GPT about a certain situation, and it can answer the question and provide information. But it may also provide a link to another source, and that source could be entirely Russian and disseminate fake information,” the analyst warned.
What tools are available to counter such information influence? Media trainer and co-founder of StopFake.org Ms. Olga Yurkova believes that fact-checking remains one of them.
“Of course, it’s not some magic pill that will solve all problems. In today’s complex world, that’s unrealistic. But let’s take a look at how Russian information operations have changed overall. Many things haven’t actually changed. Many approaches have remained the same as before. They’ve simply taken on a new scale and acquired new tools. Whereas dozens of people worked in troll factories in 2014, now all of this is done automatically. A single operator can generate hundreds of texts in different languages very quickly, in real time,” noted Olga Yurkova.
The level of information acceptance depends largely on public trust in government institutions and the media. As an example, Daria Opryshko cited a study by her colleagues on the election campaign in Denmark.
“Why didn’t the information campaigns work there? There weren’t any ‘information ghosts’ to stir up. That’s because there is a very high level of trust in society, both in the government and in the state as a whole. In such a scenario, trust will come to the forefront,” explained the expert from the NGO Human Rights Platform.
The event took place as part of the United by Truth project, supported by the Canada-Ukraine Foundation.