US MIDTERMS
The first AI election is here
Early voting is underway for the US Midterms, and it’s already making history. Because these are the first elections where AI is built into every major social platform.
During the 2024 Presidential race, the tools were much more basic. Think obvious slop, that was mockable and meme-able. But two years later, the tech has leveled up. Now, AI attack ads feature imperceptible deepfakes to smear opponents and mislead voters.
When it comes to policing misinformation, platforms are leaning on AI for that too. Social media companies have swapped human moderators for automated systems to screen posts at scale.
And if you try to bypass the social media slop by heading to search engines, you run into a new set of challenges.
You can do your due diligence ahead of the polls and still land on Google's AI overview before seeing any vetted sources or political websites. This may be convenient, but it's not inherently credible.
That’s because some of the "news" sites out there are partisan operations in disguise. NewsGuard, a watchdog that tracks harmful content online, has counted more than 1,100 of them. They call these sites "pink slime." And when NewsGuard asked chatbots about their stories, the bots cited those partisan sources nearly half the time.
The AI election, by the numbers:
1 in 5: The number of voters who plan to use AI for race research.
37: Amount of AI-filled "news" sites built in a single day to feed chatbots.
1%: Voters calling AI their most credible election source.
Don’t take AI’s word for it
We have to find new ways to work with AI, and part of that is taking responsibility for how we digest information. Take the time to see who’s behind the stories you're reading and click through your election office website before voting. Sure, it’s a bit more legwork. But it beats letting a machine decide for you.
Fact-check the bot before you check the box.
“Pink Slime” sounds too cute for how serious this is. -TL