Experts say voters cannot rely on visual clues alone to spot fake election content.
WASHINGTON —
Political content made with artificial intelligence is becoming more common and more sophisticated.
Experts say voters need to stay on guard this election season, given the lax regulations governing the use of this rapidly evolving technology.
The Associated Press reported on Wednesday that social media videos believed to be made with AI and linked to Russia are misusing real people’s names and photographs to spread disinformation about two Democratic Senate candidates in close races: Sherrod Brown in Ohio and James Talarico in Texas.
Ari Lightman, a digital media and marketing professor at Carnegie Mellon University, said foreign adversaries are using increasingly advanced tools to power a familiar playbook.
“In order to achieve a purpose, which is really to create greater levels of polarization and lead to some sort of political outcome that’s favorable to them,” Lightman said.
Political groups in the United States are also increasingly leveraging AI in campaign ads.
The Wesleyan Media Project has tracked 220 midterm election ads that were generated — at least in part — using AI tools. The analysis described the total released earlier this month as “almost certainly an undercount.”
The ads were put out by stakeholders in both parties, but most stemmed from Republican candidates and organizations.
The analysis also identified a surge of spending on AI-generated ads. The price tag had exceeded $137 million as of earlier this month.
“For comparison, we identified less than $50,000 in spending on AI-generated ads in the 2024 cycle,” said Erika Franklin Fowler, co-director of the Wesleyan Media Project, in a press release. “Much of the AI use is not disclosed through any type of disclaimer.”
According to the National Conference of State Legislatures, 31 states have enacted laws regulating the use of deepfakes in political messaging, though some have been challenged in court on free speech grounds. There are still no disclosure requirements at the federal level.
Meanwhile, it’s getting increasingly difficult to spot visual clues as AI improves.
How voters can remain vigilant
While there may be giveaways in some cases, like misshapen hands, impossibly smooth skin, or misaligned shadows, experts say voters cannot rely on these tells alone.
“These have all gotten so sophisticated. It’s really hard for the layperson to tell whether the information that’s being sent to them is real or completely fabricated,” Lightman said.
Lightman said voters should look into credentials and context behind potential AI content, and verify information through independent sources.
Voters should also be skeptical of content that seems designed to evoke an immediate emotional response and think twice before sharing on social media.
Are AI models taking action?
OpenAI confirmed on Wednesday that it’s launching an Election Operations Center this week. It will run until all votes are counted to identify the spread of misinformation and combat the potential abuse of tools like ChatGPT. The plans were first reported by Politico.
A new Brennan Center report found that OpenAI was the only company that significantly improved protections against generating fake election content since the center ran tests on six of the most popular models earlier this year.
