Lawmakers want EAC to develop guidance on AI’s risk to elections

Bipartisan legislation introduced on Friday by a quartet of House lawmakers would push the nation’s top election administration agency to develop voluntary guidelines that address the potential impact of artificial intelligence technologies on the voting process. 

The bill would require that the Election Assistance Commission — or EAC — release public recommendations around the use of AI tools on the administration of elections. The legislation was introduced by Brian Fitzpatrick, R-Pa., Chrissy Houlahan, D-Pa., Doug Lamborn, R-Colo., and Abigail Spanberger, D-Va.

According to the proposal, the EAC’s voluntary guidelines should address “the risks and benefits” of AI on election administration; the cybersecurity risks of these technologies; how AI-generated and shared content can affect the sharing of accurate election information; and how AI-generated disinformation can undermine trust in the voting process.

The recommendations would be developed in consultation with the National Institute of Standards and Technology and would be issued to state and local election offices and Congress, in addition to being made publicly available. The EAC would have 60 days from the bill’s enactment to release the guidelines. 

“With the rise of artificial intelligence, we must equip our election administrators with the necessary tools and guidelines to safeguard our democratic process, Houlahan said in a statement, adding that the bill “underscores our commitment to fair and secure elections by addressing the potential risks posed by AI, including misinformation dissemination and cybersecurity threats.”

The legislation is the House companion to a Senate measure introduced in March by Sens. Amy Klobuchar, D-Minn, and Susan Collins, R-Maine.

In a March 12 statement, Klobuchar — who chairs the Senate Rules Committee — said the bill “will prepare state and local officials to address the risks that AI poses to our elections.”

“We know that AI is being used to spread disinformation about voting,” she added. “To safeguard our free and fair elections and support hardworking election officials, comprehensive guidelines are needed to address AI’s impact on election administration.”

AI-generated content and disinformation have already begun to have an impact on the 2024 U.S. election process, although some agencies have moved to mitigate the risk posed by the technologies. 

Ahead of New Hampshire’s presidential primary in January, residents received a robocall with an AI-generated voice of President Joe Biden that told voters not to head to the polls. Following outcry over the incident, the Federal Communications Commission issued a unanimous ruling in February that deemed it illegal for robocalls to use AI-generated voices. 

That same month, the EAC unanimously voted to allow state election officials to use federal funds to “counter foreign influence in elections, election disinformation and potential manipulation of information on voting systems and/or voting procedures disseminated and amplified by AI technologies.”