OpenAI launches election operations center to monitor misinformation

The Election Operations Center

OpenAI has announced the creation of a dedicated Election Operations Center to oversee the 2026 midterm elections. This initiative, which some are describing as a 'war room,' will remain active until every vote in the midterm cycle has been counted. The primary goal of the center is to identify and mitigate the spread of misinformation and other forms of election disruption.

The company intends to use this centralized hub to bring different teams together for a unified response to emerging issues. By consolidating resources, OpenAI aims to enforce its existing policies more effectively as the public heads to the polls. This move follows years of work by the company to address civic risk and strategic threats posed by artificial intelligence.

Josh Lawson, who leads civic and strategic risk at OpenAI, stated that the center is designed to respond quickly to new problems. He noted that the company has put in considerable effort to ensure users can access reliable voting data. The center serves as a proactive defense against the misuse of AI tools during a high-stakes period of civic engagement.

Combating deepfakes and deception

A significant part of the new operation involves preventing the creation of deceptive media. OpenAI is developing safeguards to stop 'bad actors' from using the system to distribute content to large audiences more quickly. This includes specific measures to handle deepfakes, which are highly realistic but fake images or videos created by AI.

One specific technique mentioned is 'cartoonification.' This process involves the system generating a cartoon version of a prompt rather than a photorealistic image. For example, if a user asks for a photo of long voting lines, the system would produce a cartoon instead of a realistic image to avoid creating deceptive content. This prevents the technology from being used to manufacture fake visual evidence of voter interference.

Lawson explained that these safeguards are necessary to protect the integrity of the voting process. By limiting the realism of generated images that could be used for misinformation, the company hopes to reduce the risk of visual deception. The goal is to maintain a balance between creative utility and the prevention of photorealistic voter interference.

Partnerships for reliable data

To ensure that the information provided by ChatGPT is accurate, OpenAI is entering into several key partnerships. The company has teamed up with The Associated Press and Democracy Works to source helpful and reliable information for its users. These organizations will provide the foundational data used to answer questions about the voting process.

One of the most notable aspects of this partnership is how election results will be handled. ChatGPT will specifically use The Associated Press as its source for calling elections. This ensures that the information presented to users regarding winners and losers comes from a recognized and established news organization.

The collaboration with Democracy Works also focuses on providing nonpartisan resources. By using data from these trusted partners, OpenAI aims to help users navigate the complexities of voter registration and polling locations. These partnerships are intended to ground the AI's responses in verified facts rather than generated guesses.

The broader AI landscape

OpenAI is not the only major AI player taking action during the election cycle. Anthropic, the maker of Claude.ai, is also working to battle misinformation surrounding the election. However, Anthropic noted that the majority of its users do not typically use the platform for political discussions or social issues.

Data from Anthropic shows that only 5.75% of all chats on Claude.ai touch on politics or social issues. Furthermore, only 0.51% of conversations involve specific questions about elections or voting. The company reported that asking who to vote for accounts for just 0.01% of all conversations on the platform.

Despite these low numbers, Anthropic has still rolled out specific safeguards for the election. The company added a banner that links to TurboVote, a nonpartisan resource from Democracy Works, when users ask about polling places or registration dates. This mirrors the broader trend of tech companies, like Meta, which have historically established election centers to manage civic integrity.