Electoral disinformation: limited impact of generative AI in Canada

The recent federal electoral campaign was the first to take place in the era of generative artificial intelligence (AI). Despite the initial fears, this technology has not deceived voters much since it has been mainly used for humorous or political communication, according to experts consulted by Decryptors.

Listen, friends, I see all those messages that say that I am the ideal candidate for Canada. But what are you talking about? I spent more time abroad than in Canadian soildeclares a hyperture version of Prime Minister Mark Carney in a video posted on Tiktok the day before the ballot. I do not represent the people, I represent the elites.

This video, clearly labeled as having been generated by IA By this platform, one of several similar contents identified by the Observatory of the Media ecosystem of McGill and the University of Toronto. His goal was not to deceive.

People know that this video is not true, but it is representative of reality for some peopleexplains the director of the observatory, Aengus Bridgman. Yes, we saw a little mesination, but most of the content generated by IA that we have seen was used for humor or satire.

This does not mean that this type of hypertructure, comparable to caricatures, does not mistaken anyone.

Another rigged video posted online Thursday by this same account shows Prime Minister Carney who announces the entry into force of a new carbon tax although he abolished the previous one in mid-March. Most users in the commentary section capture the humorous intention of its creator, but others do not seem to understand the joke. She was seen some 380,000 times in a day.

Other images of a more caricatured nature which showed for example Mr. Carney disguised as superhero in the colors of the flag of China or the conservative chief Pierre Hairyvre in Captain Canada Sitting on a polar bear also circulated on X.

Open in full screen

Many Tiktok accounts have used generative AI tools to create content. They were not necessarily very followed and did not receive a lot of attention, but their content was still broadcast on the pages “for you” from users throughout the campaign. It gave a certain impression of a polluted information environment, even if the public was not necessarily deceived by the contentobserves the director of the Pol Comm Tech lab at the University of Ottawa, Elizabeth Dubois.

Images IA Virals of Carney and Epstein

This principle of illustrating ideas using theIA was particularly observed at the start of the campaign, when images showing Mark Carney with the financier and sexual criminal Jeffrey Epstein circulated abundantly on the social network X.

Unfounded theories (new window) According to which Mark Carney had links with Jeffrey Epstein circulated on social networks in March and these photos may have played a role in their propagation, without necessarily believing in the authenticity of images.

These images led to the idea of ​​caricature or humor less. They aimed to obtain headlines in the media and attract attention rather than simply polluting the informational environment or disinforminganalyse Elizabeth Dubois.

Open in full screen

Theories were based in particular on a real photo of Mr. Carney alongside the close partner of Jeffrey Epstein, Ghislaine Maxwell, a photo that also surfaced during the campaign.

The photo in question had been taken during a British music festival in 2013. According to a survey by Toronto Star (new window)who was able to join the photographer who had captured this scene, Mr. Carney and Mrs. Maxwell had not talked that day. The former banker seemed uncomfortable And walked with his wife. At the time, he had just been appointed governor of the Bank of England and his presence at this festival was explained by family ties to the owners of the field.

A representative of the campaign of Mark Carney indicated to the Toronto Star that it was false allegations and disinformation. Another confidential source cited by this daily said that Mr. Carney and Mrs. Maxwell were not friends.

Scams and false articles doped at theIA

Throughout the campaign, fraudulent advertisements that integrated the image of federal party leaders were broadcast on several platforms, including those of Meta. They led to false articles which usurped the institutional image of known media as Radio-Canada in order to convince people to invest in cryptocurrency scams.

These fraudulent advertisements have been the subject of investigations of several media over the years, including Radio-Canada, and continue to circulate after the electoral campaign. Their shape changes according to the news.

In many cases, the advertisements in question use hypertructures of politicians like Mark Carney and Pierre Hairy, which announce an alleged new investment opportunity for all Canadians. They can use false images of politicians who are arrested by the authorities.

Open in full screen

These are scams, but the content of the articles is political disinformationEspest Alengus Bridgman. This case shows very well that the platforms are really not interested or able to manage political content well during an electoral campaign.

Reassuring … for now

These two experts agree that the situation was less serious than what several observers and experts apprehensive during the last electoral campaign. This does not mean, however, that they are not worried about the future.

We have seen a lot of experimentation and many people who have tested these tools in different ways. I expect that, by our next federal election, approaches become more sophisticated as technology continues to improvedit Elizabeth Dubois.

Since we expect to see more and more content generated by IA on social networks, it could be that people are increasingly suspicious of what they will see on the Internet and that it is more and more difficult to distinguish the true from the falseEspest Alengus Bridgman.