Elections Canada has launched an online tool to tackle “misconceptions” related to the election process as some are warning that fake content and inaccurate information could be rampant in 2024 and coming years due to the use of artificial intelligence.
The online tool, called Electofacts, is meant as a resource for Canadians to check whether information they find on the electoral process is true or not, the webpage states.
It says studies have shown that there is false or misleading election-related content on social media and links to a McGill University study, but notes that Elections Canada is “not the arbiter of truth” and does not monitor the accuracy of content posted on social media by politicians.
“We can, however, provide correct information about the administration of federal elections,” the webpage reads.
“We are doing this because it is important; Canadians should have easily accessible and accurate information about the federal electoral process directly from the source.”
In an email to Global News, Elections Canada said the goal of the webpage is to be a useful resource for Canadians.
The tool lists “narratives” that it says are some of the most prevalent misconceptions observed by Elections Canada.
Examples of those include the claim that Elections Canada lost, ignored or refused 205,000 mail-in ballots during the 2021 election, to which Elections Canada said, “Elections Canada counted all the mail-in ballots (special ballots) that it received by the election day statutory deadline and that were not spoiled.
“We cannot count special ballots that are never sent back to us or are returned after election day. This is clearly stated in the Canada Elections Act. By law, we cannot even open the envelopes that contain the special ballots that come in after the deadlines set in the Act. We must keep the envelopes, unopened, for 10 years after the election.”
The portal also addressed claims like those suggesting the vote-by-mail system allows people to vote more than once (“An elector who applies to vote by mail cannot vote at advance polls or on election day”); that election results are rigged during the ballot-counting process (“Counting votes during a federal election is always done by two election workers. They count each ballot in front of observers, including candidates or their representatives”); and the deceased receive voter cards that could be used by others to vote (“Under the Canada Elections Act, it is an offence to pretend that you’re someone else in order to vote. Every elector must provide proof of their identity and address before being allowed to vote”).
Each of these claims is paired with points identifying the rules as laid out under the Canada Elections Act and other publicly available information about the mandate and role that Elections Canada plays in administering elections.
Elections Canada’s tool comes as some experts warn that AI could have an impact on elections in 2024. The Communications Security Establishment (CSE) stated last year that AI-created deepfakes — computer-generated images and videos that show events that did not take place — “will almost certainly become more difficult to detect, making it harder for Canadians to trust online information about politicians or elections.”
While the technology isn’t widely used yet, the report states the CSE expects use to increase in the next two years.
University of Waterloo assistant professor Mark Crowley, who specializes in machine learning, told Global News that the societal issues of AI will dominate 2024, and one concern is the generation of fake news by the technology, such as by generating images or voices, which Crowley said is much more readily available than it was before.
“You really have to be wary whether the people that you’re seeing talking on social media are real people or not,” he said. “It’d be surprising if (AI content) didn’t have some impact.”
The CSE report raised similar warnings, with CSE chief Caroline Xavier saying that “AI-generated deepfakes can create a video of something that never happened and put words into politicians’ mouths that they never said.”
Crowley said now anyone can generate AI-created fake content, whereas in previous years, like during the 2016 U.S. election, the concern was more around content farms doing it. Crowley said the volatile political environment in the U.S. now makes AI content more worrisome since it could be used to push false stories and undermine reasoned debate.
Adding to the storm of fake content in 2024 could be AI-generated video, according to University of Toronto associate professor Colin Raffel, who also specializes in machine learning.
He expects video generation by AI to make significant progress in the next year after image and text generation took off in 2023.
He said that 2023 was important in highlighting the risks of AI, but it is still something society is not even close to being on top of.
Comments