You might think that deepfakes would have a big effect on elections, but it turns out they’re not that good yet. At least, that’s what some recent research says. Despite all the fake images floating around the internet, it looks like people can still spot a phony when it counts.

Researchers found that during the global elections in 2024, AI-generated misinformation didn’t have a significant impact. Only 27 pieces of fake content went viral during the European elections, according to a report from the Alan Turing Institute. Most people who saw these fakes were already sympathetic to the messages, so it didn't really change their minds.

For example, there was an image of Kamala Harris standing in front of Soviet flags. It was clear to many that the image was fake, thanks to some obvious signs like a weird arm length and faces not reflecting right on mirrors. People on social media were actually more likely to mistake real images as fakes, showing they’re pretty good at spotting BS.

In the U. S. , out of over 1, 000 pieces of misinformation about the presidential election, only 6% were made using AI. Most talk about deepfakes online happened when new generative models were released, not during election time.

Even though AI imagery isn’t perfect, it’s still a concern. Fake media can still reinforce beliefs even if people know it’s not real. Plus, the confusion around real and fake media can hurt trust in online sources. AI has even been used to create harmful deepfakes targeting female politicians.

As the technology gets better, we need to stay alert. It might not be fooling everyone now, but that could change.

Fake Videos Had a Minimal Impact on Elections

Actions