We haven't seen too many AI shenanigans in our elections so far, but that can always change.

From American Public Media, this is Marketplace Tech. I'm Megan McCarty Carrino. The 2024 presidential race was the first big election to happen in the new generative AI era.

There have, of course, been major concerns that these tools could be used to deceive voters

or interfere with the exercise of democracy.

But so far, that kind of activity has been limited,

according to a new report from the Center for Democracy and Technology.

Tim Harper is a senior policy analyst and co-author of that report.

Campaigns mostly used AI for speed and for scale, but not for spectacle.

Consultants told us AI's biggest value was generating rapid, targeted messaging across multiple audiences, not cranking out deepfakes.

The real AI revolution in politics isn't a fake video.

It's faster persuasion.

It's faster analyzing data. It's creating emails, doing outreach, building strategy for how to manage their staff, their engagement,

their timeline. Why do you think AI didn't play kind of a bigger role or, you know, what held off the worst predictions about how AI might influence our elections?

Well, I think that there were a number of voluntary guardrails that were put in place.

Mostly these were norms. Campaigns were worried that voters would penalize them at the polls if they

created manipulated videos. And I think that that led people to be pretty conservative with

how they used these tools in 2024 as the norms were still being developed and catching up.

I think the problem is that as those

voluntary norms and expectations begin to erode, campaigns will start to see the risk to reward

change. If voters are more used to seeing campaigns create manipulated videos, they're more likely to think

that, well, I need to do that to keep up, to be competitive against my opponents.

And so that kind of switches the motivation structure so that rather than being constrained by the

...