Anthropic To Feed Claude Official Voting Data Ahead Of November Midterms

Authored by Kimberly Hayek via The Epoch Times,

Anthropic announced Wednesday that it will add official U.S. voting information and election results supplied by the nonprofit Democracy Works to its Claude chatbot ahead of the Nov. 3 midterm elections.

Democracy Works provides “Americans with the guidance they need to vote in every election, every time.” The nonprofit first partnered with Anthropic in 2024 “to prevent the misuse of AI in political contexts.” In April 2026, Anthropic began showing Democracy Works’ nonpartisan TurboVote tool in a banner when Claude was asked about voting registration, locations, or dates.

Later this month, Claude will start drawing official voting information from Democracy Works into its responses on Claude.ai for users in the United States, Anthropic said.

“Someone asking where to vote or what ID to bring to their polling place will now get an answer specific to their state, without leaving the conversation,” the company said. “These responses draw on publicly available information that Democracy Works collects from state and local election sources.”

Claude will also begin showing up-to-date results on election night and in the days that follow.

“When people ask for current vote counts or election results, they’ll be shown the latest official numbers directly in their conversation,” Anthropic said. “Our goal is for Claude to be a useful tool for political conversations, while preventing election-related abuse.”

Conversations that touch on politics, public policy, or social issues make up about 5.75 percent of all U.S. chats on Claude.ai, Anthropic reported after reviewing 250,000 conversations from late September. About 0.51 percent focused on elections or voting. Requests for advice on whom or what to support appeared in just 0.01 percent of those exchanges. Policy violations were rarer still, at roughly 0.002 percent. Most requests stay informational, Anthropic said, with people asking where candidates stand on issues, weighing policy arguments, or drafting documents.

The company expects political queries to climb as Election Day nears.

“More people are turning to Claude for information about this year’s elections than ever before,” Anthropic said. “We have a responsibility to make sure the answers they get are accurate and balanced, and that Claude isn’t used to undermine the democratic process.”

Claude Opus 5.5, Sonnet 5.5, and Haiku 5.5 scored 99 percent on the firm’s political evenhandedness evaluation – the highest to date.

“When people ask Claude about politics, they should get accurate and evenhanded answers that don’t favor any particular political viewpoint,” the company said.

Independent reviews by the Future of Free Speech, the Foundation for American Innovation, and the Collective Intelligence Project found the models rarely refuse political requests but sometimes insert unsolicited caveats.

Anthropic added new detection tools that look for harmful activity spread across several steps, such as building fake news sites or running schemes that appear ordinary when examined one piece at a time. Tests by the Institute for Strategic Dialogue and the McCain Institute found that Claude stops single, clearly harmful requests but can miss coordinated efforts broken into smaller, harmless-looking tasks.

“Building on these findings and our own, we’ve added new classifiers to better identify novel and multi-step misuse,” Anthropic said.

AI-powered chatbots and search engines are increasingly used to find and summarize election information online, but they are not official sources and frequently cite incorrect information. Anthropic said it will keep testing for political evenhandedness and misuse as the cycle intensifies.

“We’ll continue to invest in making Claude more evenhanded, preventing and addressing misuse, and being transparent in our actions,” the company said. “This work will be critical in the upcoming U.S. midterms and extends to elections around the world as well.”

Anthropic has faced separate scrutiny this year over model access, export controls, and government contracts, though those disputes sit outside its voting information plan.