The last session of today’s e-Society.mk International Conference, which has been organized by Metamorphosis Foundation for the 19th time this year, was dedicated to the issue of disinformation and abuse of artificial intelligence for malign influence on democracy or, more specifically, on the elections and the opinions of the voters.
The topic “Disinformation on Steroids: Threat to Election Integrity” was covered by: Jan Cingel, founder and president of Strategic Analysis from Slovakia, Bojana Kostić, Digital rights researcher and advocate from the Netherlands; Fitim Gashi, Project manager against disinformation in Sbunker from Kosovo, as well as Filip Stojanovski, Partnership and resource development director in the Metamorphosis Foundation for Internet and Society.
First, Stojanovski presented the Western Balkans Anti-Disinformation Hub project, initiated by the Metamorphosis Foundation and implemented jointly by six Western Balkans organizations in an effort to subdue malign influences of disinformation. Due to the linguistic familiarity and proximity, these influences cross the Balkan borders very easily. Stojanovski presented some examples through texts debunked by fact-checkers from the six organizations.
Jan Cingel presented the Slovak experience with a harmful influence on elections. The elections in Slovakia took place last September after being run by a technical government for a specific period, and they were the first instance of using artificial intelligence to influence public opinion.
“Apart from the presence of pro-Russian narratives that mainly spread on social media, we also had two examples of artificial intelligence abuse by creating deepfakes that although low-quality it was still obvious that they were a product of artificial intelligence. Nevertheless, they managed to mislead many people who believed they were true and influenced a large number of voters”, explained Cingel.
The first deepfake example is a fake audio file created with artificial intelligence, low in quality and not taken seriously. This artificial intelligence product was a “conversation” of the President of the party Progressive Slovakia, who allegedly was saying that if they won, they will increase the price of beer. Although this issue is quite sensitive in a beer-drinking country, they still managed to deceive many people, he explained. The second example, however, had a much stronger effect.
“The second video created with artificial intelligence was more sophisticated. It presented an alleged conversation of the President of the Liberal Party with an investigative journalist covering corruption scandals of the previous government. In the conversation, the alleged party leader was devising plans for manipulating the elections – how to buy votes from the Roma population, how to buy journalists and other similar topics. The quality of this video was low, but still started to spread mainly on the Telegram channels of pro-Russian public figures. After that, the video hit the social media”, explained Cingel.
He stressed that the effect of the fraud was greater with the right-wing voters.
“Those people who wanted to vote for the right and the far-right said that the liberals were trying to steal the elections. The deepfake confirmed their suspicions, but did not change the opinion of many people who wanted to vote for the liberals”, said Cingel.
Fatim Gashi from “Sbunker” spoke about the issue of influencing the political situation in Kosovo by means of disinformation, including some “more scandalous” information in the Albanian language area, while expert Bojana Kostic covered the measures undertaken by the EU so far to protect digital rights and the fight against disinformation, as well as the measures that need to be implemented to prevent abuse of artificial intelligence.
