OpenAI Outlines Election Safeguards Ahead of Midterms

OpenAI on Wednesday laid out a broad plan to combat election misinformation and cyber threats ahead of the 2026 midterms, announcing new partnerships and backing legislation that would create new rules around the use of artificial intelligence in elections.

The efforts, which were first reported by Axios, effectively serve as an acknowledgment that voters and political consultants are increasingly turning to AI for news and information about politics and that the emerging technology has the power to influence the course of elections in the U.S. and abroad.

New polling conducted by the Rainey Center found that around 1 in 5 voters has either already used AI tools to learn about politics or make voting decisions or said they might do so in the future. Among those voters, 44 percent said that they would use AI to summarize news or political topics, while 43 percent said that they would use AI tools to fact check claims.

But despite the growing popularity and adoption of AI tools, there are lingering concerns about the reliability and trustworthiness of AI-generated political content. The Rainey Center poll found that, when it comes to AI in politics, 62 percent of voters are most concerned about “false or misleading content.”

In an effort to direct users to reliable sources of election information, OpenAI is partnering with the Associated Press this fall to provide live vote counts on election night in the U.S. and Brazil. The company is also teaming up with Democracy Works to display logistical information on voter registration processes and precinct locations. 

OpenAI announced that it would also offer cybersecurity products to registered voting system manufacturers and would engage the National Association of Secretaries of State and the National Association of State Election Directors “to ensure state election authorities are briefed on the latest cyber capabilities and our tools for defenders.” 

The outlined election plan also details an effort to crack down on deepfakes using a dual approach: SynthID watermarks, which embed an invisible watermarking layer onto images generated by ChatGPT, and the enforcement of standards that use metadata and cryptographic signatures to help information about an image travel with the content itself.

“People are increasingly using AI tools to create content they then share on social media, messaging apps, and the web,” the company wrote in a blog post. “To help combat misleading ‘deepfakes’, we are investing in a multi-layered provenance approach that will equip people to verify whether content they’re seeing has been created or modified with AI.” 

OpenAI said it prohibits using its tools for election interference, voter demobilization or deceptive campaigning, claiming its detection and enforcement systems have improved since 2024. Its recently launched advertising business will not run political advertising this cycle, the company announced. 

The company is also endorsing a pair of bills: the Protect Elections From Deceptive AI Act, which would ban the distribution of materially deceptive AI-generated audio or visual media related to federal candidates, and the Preparing Election Administrators for AI Act, which would task the Election Assistance Commission to create voluntary guidelines for election offices on the use and risks of AI technologies in election administration.

“We will continue to learn from partners, update our safeguards, and adapt,” OpenAI wrote in its blog post. “Our goal is to support people’s ability to participate freely in elections and make their own decisions — with reliable information, transparency, and effective safeguards.”