How A Deepfake Might Have Cost This Candidate An Election

We’ve all heard about the potential dangers associated with artificial intelligence … and so-called deepfakes in particular.

Some misleading uses of manufactured audio and video might not be all that serious (like the recent fake clip of Tom Hanks shilling for a dental plan), but others can cause untold damage.

Slovakian scandal

With just two days left until a major election, an audio clip surfaced on social media that seemed to reveal a plot to commit voter fraud. The voices were supposedly from candidate Michal Simecka and journalist Monika Todova and the topic of discussion involved paying members of the nation’s Roma community to vote a certain way.

Although Simecka and Todova asserted that the audio was fake, there was a clear limit to their ability to get the word out or stop the spread of the deepfake on Facebook.

Here’s how it played out:

• The audio surfaced during a two-day moratorium on election-related speech by politicians and news outlets.
• Since it was just audio, the post didn’t violate Meta’s policy against sharing manipulated video clips.
• After the election, news surfaced that Simecka’s party lost the election to a rival party that wants to pull out of Ukraine.

This election was already seen as an example of how European elections might be impacted by rapid technological advancements, and it appears to have highlighted a serious vulnerability.

Action required

Veronika Hincova Frankovska of the fact-checking group Demagog explained that this incident proves democracies are “not as ready for it as we should be.”

Part of the problem lies with the ability for AI-generated misinformation to spread across social media, but Meta spokesperson Ben Walter defended the company’s policy of labeling and down-ranking fact-checked content.

As recent developments in Slovakia demonstrate, however, that policy no longer seems to be sufficient.

Chris Agee October 4th, 2023

Share this story: