How Russian bots started to divide Ukrainians and Crimean Tatars

Offensive comments, dividing Ukrainians and "Tatars" and aggressively emotional statements against Islam. Dozens of such comments began to appear on social networks under posts of Ukrainian media outlets talking about Crimea. At first glance, it may seem that discussions using hate speech are led by real people. However, most of them are Russian bots that were involved in a campaign to discredit the Crimean Tatars, says Yulia Dukach, head of disinformation research at OpenMinds.
"If we see very aggressive hate speech from a user we don't know, then it is very likely that it is a Russian bot," explains Yulia.
We noticed a massive appearance of offensive posts, especially under publications Сrimea Vox, Suspilne Krym and Qirim media. Here, for example, are excerpts from the comments below Crimean Vox's Instagram post about the start of Crimean-Tatar language teaching at the Kyiv-Mohyla Academy: "The Tatar-Mongols have officially occupied Kiev!", "As soon as even a small part of this contagion called Islam reaches Ukraine, it will be extremely difficult to eradicate it", "Tatars are not even a Ukrainian ethnic group, but a foreign one, with a different culture, religion and language", "The Mogila surfaced. Find out when Tatars appeared in Crimea, and when Ukrainian-Rusyns lived there".
Ethnic hatred or religious intolerance is not typical for Ukrainians, explains Julija Dukač. At the same time, Russia could launch such a discriminatory campaign as part of an experiment: whether it will succeed or not.
"If this topic succeeds in Europe, will it suddenly succeed in Ukraine? And Russia's main strategy is to use all possible means and topics that can sow divisions and can enable Russia to become stronger, because Ukraine will become weaker," explains the expert.

Crimean Tatars are always referred to in the comments as “Tatars”, without the word Crimean. Although Crimean Tatars, the indigenous people of Crimea, and Tatars living in the Russian Republic of Tatarstan, are two completely different ethnic groups, Yulia Dukach believes that the goal of the Russian campaign here is precisely the separation of the indigenous people from Crimea. Here is another comment from the bot under the post on the Crimea Vox page on Instagram: “The time will come, our people will conquer Crimea. Then the Crimean Tatars will come and say: bravo, bravo. Now we declare Crimea independent, thank you for conquering it, you bastards”.
And this is an illustration of another narrative promoted by Russian bots: the separation of Ukrainians and Crimean Tatars.
“Here the image of “us” and “them” is created. Ours will come, they will liberate, and then the Crimean Tatars will come. That is, the Crimean Tatars are not “ours”. This is a separation from Ukrainians, although we are talking about a political nation that may include representatives of different ethnic groups and religions. This is not something we can oppose,” explains Yulia
The artificial opposition in this Russian bot comment is reinforced by the emotional hook about the liberation of Crimea, says the expert:
“That is, a person feeds on what he wants to read or hear. And if an account with a patriotic avatar writes it, you can’t tell it’s a Russian bot. Because how will a Russian bot write about the liberation of Crimea? They have learned to be more stealthy, they are not afraid to write in Ukraine, not in Ukraine, even when they write in Russian.”
Yulia Dukach says that Russian bots, with the help of artificial intelligence, look very similar to people: these accounts have names, surnames, often Ukrainian, profile photos and even a history of posts. Simulating a virtual person who looks like a real person using artificial intelligence is now not a problem:
“Currently, generative artificial intelligence makes it very cheap. Previously, Russian bots stole other people's photos, for example, from Instagram and created a profile on the former Twitter, now X. Now one sentence in a conditional GPT chat and you have a photo that looks very similar to a person who is not actually a person,” says Yulia.
And she adds that Russian bots usually have extremely patriotic Ukrainian avatars and photos. To estimate the size of the Russian bot farm, experts from OpenMinds analyze posts on social networks. 

“To prove that it is a bot, we need to see that the same user left comments on the same topic on 50 different channels. I did some research, and so far this was the largest bot farm, which was 3,600 accounts that were working in Telegram comments in the territories occupied since 2014. These accounts left 310,000 comments in a year and three months. And when we looked at where else they were writing and saw that these same bots were active in both Russia and Ukraine during this same time period, we saw a figure of 3.5 million comments left by 3,500 accounts,” the expert explains.
This study also showed that the same bots leave negative comments under posts in Ukraine, while under posts in the territories occupied by Russia they leave positive ones. For example, they praise the occupation authorities.
“There was a special cluster about Crimea, that ‘Crimea is Russia.’” They also congratulated May 9, wrote that it was good that the "authorities" introduced a curfew, that the "authorities" were concerned about our safety. More than half of the comments in the occupied territories were pro-Russian, while the same bots in Ukraine were pro-Russian, writing about the puppet government, about the “Kiev regime” and so on,” says Yulia.
According to her, the bot system works according to an algorithm and, in fact, monitors the entire Ukrainian information space. If the algorithm contains keywords, for example, the Crimean Tatar language or Crimean Tatar, the system conditionally gives a signal to the bots and the bots come up with prepared topics.
“They have a narrative that they have to spread. For example, hatred of Muslims. The narrative that there is no place for Muslims in Ukraine. What are they doing at this moment? They look at the post, the artificial intelligence model analyzes the text and generates a comment that will look as relevant as possible in that context. That is, they act as quickly as possible without human intervention, because it is an algorithm. And if they have already started doing this, then most likely there will be more and more different comments under the next videos, because they will not be repeated like before, the tone may be different,” explains Yulia.
Yulia Dukach says that Russian bots are being adapted with the help of artificial intelligence to make them harder to recognize, that is, to make their comments as similar as possible to those written by real users.
“That is, they write in small letters, use a more formal tone, add errors to their text, sometimes these errors are the result of the work of artificial intelligence, which does not write perfectly in Ukrainian. Sometimes they add some emotionally colored words, swear words, and then we no longer notice that there is no discussion, but we react to some emotional hook.”
According to experts, conducting a discussion with bots is not productive. Firstly, from the point of view of wasting time, and secondly, when a person enters into a discussion with a bot, it, in fact, gives him a platform to repeat his narratives once again.