Voters are using AI to learn about candidates, the issues, current events, and how to vote. Do you know what it’s telling them? In 2024, just 11% of voters said they used AI chatbots to get election information, but with the rise of adoption and increases in zero-click searches on Google, we can expect that number to be far greater in 2026.
Your campaign needs to be on top of how ChatGPT, Claude, Grok, and Gemini are answering questions about your candidate and the race. Here’s how you can do this for free with a little bit of time.
How are you really using AI?
We're partnering with Anchor Change, Chaotic Era, and Doomscroll on a survey to better understand how political professionals like you are using AI.
Key Questions Voters Ask
Start with a list of five key questions that voters might use when researching your campaign. Copy and paste each one of these into the chat and fill in the blanks:
- “Tell me about [candidate name]” Does the AI know you exist and what does it lead with? Learn what AI sees as the defining facts about your candidate.
- “Who is running for [office] in [state/district] in 2026?” Can a voter find you if they don’t already know your name? If you’re not in the list of contenders, find the sources it’s pulling from and get yourself added.
- “What are [candidate name]’s positions on the major issues?” Find out if the AI is misrepresenting any of your positions with outdated, partial, or incorrect policy positions.
- “Why should someone vote for [candidate name]?” See how your campaign’s message is or isn’t breaking through to AI.
- “What is [candidate name]’s background and experience?” Learn whether or not your biographical information is being presented correctly.
- “What is [candidate name]’s voting record?” If your candidate is an incumbent, find out how it frames past votes.
- “[candidate name] vs [opponent name]” This gives you the clear head to head matchup to see what AI says about both candidates.
Where & How To Test
Test these prompts across the most widely used AIs: ChatGPT (OpenAI), Claude (Anthropic), Gemini (Google), and Grok (xAI). These are the tools voters have the most exposure to.
Make sure you have web search enabled for all of these because it will more accurately mirror how voters actually use them. For even more insight, you can turn off web search to see what information is baked into the model already.
Repeat The Test
If you ask an AI the same question twice, you’ll get two different answers, so a single check tells you what the model said once, not what it typically says. Run each prompt at least twice per platform. You’re looking for the patterns that emerge with each one.
Start with a new conversation each time. Continuing an old thread will mean that prior messages will influence subsequent responses. Make sure you have memory turned off (if applicable) before running these tests.
What To Look For
You’re looking for four things with each response:
- Is your candidate mentioned at all?
- Are the facts accurate?
- What framing does the AI use?
- What sources does the AI cite?
Based on the sources it cites, you may have some work to do adding information (like on your campaign website and social media) or working with Wikipedia and Ballotpedia to update it.
Conclusion
This isn’t a one time audit, either. The AI outputs will always be changing so it’s worth repeating, especially after major events in the campaign.
Voters are already using AI to inform their decisions and your campaign needs to study these platforms carefully.