Bipartisan lawmakers press agencies on AI election threats

A bipartisan pair of House lawmakers are pressing multiple federal agencies over the risks artificial intelligence could pose to the upcoming election, specifically over chatbots’ responses to voters.

Reps. Josh Gottheimer (D-N.J.) and Mike Lawler (R-N.Y.), in a letter sent Tuesday, urged the heads of the departments of Homeland Security and Justice, the Cybersecurity and Infrastructure Security Agency (CISA) and the Federal Election Commission (FEC) to “work together” to ensure AI models are not inconsistent when answering voters’ questions.

“As millions of Americans rely on AI-driven tools to research the upcoming election, the accuracy and neutrality of these tools are now directly tied to the integrity of our democratic process,” the lawmakers wrote.

The House members cited a 2024 study from the Institute for Advanced Study that found AI chatbots did not consistently give accurate, unbiased, harmless or complete answers after researchers submitted questions voters may ask during an election.

According to the study, more than one-third of the AI models’ test responses were marked as harmful or incomplete, and all of the tested models “performed poorly” regarding election information.

The study was completed on models last updated at the end of 2023, and models have drastically changed since then. The lawmakers reached out to the major AI firms behind these models earlier this year and said the companies pointed to the “progress” made and “limits of voluntary safeguards in addressing political bias and misinformation.”

The lawmakers also pointed to an article that claimed about 40 percent of AI “gets its facts” from Reddit, followed by Wikipedia and YouTube.

“These sources present an environment filled with unverified information, where anyone with an opinion can shape the narrative,” the letter stated. “By relying on this information, systems cannot reasonably present a full picture of unbiased information to voters.”

“With the midterm elections just a few months away, we urge your agencies to coordinate to

ensure the U.S. is prepared for AI-driven threats to election integrity, including the development

and deployment of appropriate monitoring and mitigation capabilities,” it added.

The letter was sent just days after The New York Times reported on how voters are increasingly turning to AI before casting their ballots and treating them as “a viable alternative to traditional news coverage.”

The Hill reached out to CISA and the departments of Homeland Security and Justice for comment. The FEC declined to comment.

Updated at 1:59 p.m. EDT