AI deepfakes could sway voters and disrupt elections

AI Video & Visuals


You can enable subtitles (captions) in the video player.

Though it may sound like him, this phone message sent to voters in January did not come from Joe Biden.

This Tuesday's vote will only enable Republicans to achieve their goal of re-electing Donald Trump.

Rather, it is a fake, AI-generated email urging recipients not to vote in New Hampshire's Democratic presidential primary.

It is important to save your vote for the November election.

And the foul-mouthed rant by UK Labour leader Sir Keir Starmer, who scolded party members last year, is also an audio deepfake.

I literally said, [bleep]a complete idiot.

By the end of the year, roughly two billion people will head to the polls, including in the United States, the European Union, India and the United Kingdom. Their decisions will have far-reaching effects at both the national and geopolitical levels, as elected governments implement policies that affect issues ranging from taxation and spending to trade and international relations.

But today, thanks to powerful tools like ChatGPT, it's cheaper and easier than ever to manipulate the media, and even elections. While it may be relatively easy to spot that these images of Donald Trump are fake, it's becoming increasingly difficult to spot manipulated media.

Fake audio in particular could be a powerful weapon for those trying to disrupt the election: New audio technology allows AI to replicate audio from just three seconds of recording.

In February, 20 technology companies, including Amazon, Google, Meta, Microsoft, TikTok, and OpenAI, announced they would join forces to combat the creation and spread of AI-generated content aimed at misleading voters. The effort could include adding audio and video watermarks to content to make clear its origins and whether it has been altered.

According to a study by the World Economic Forum, AI-generated misinformation is seen as one of the biggest global risks for 2024, after extreme weather.

Lawmakers around the world are fighting back: In February, Brazilian authorities banned the use of deepfake technology for electoral purposes, and in March the EU introduced provisional legislation to regulate AI models and systems to comply with certain transparency obligations and EU copyright law.

By the end of May, 44 states in the United States had introduced or passed bills to regulate deep fakes in elections. So far, there have been no major issues with deep fakes in this year's elections. However, as the amount of fake media generated by AI increases, it will become increasingly difficult for voters to believe what they are hearing and seeing.



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *