ChatGPT is adding live election information—what should voters still verify?
OpenAI says US voting guidance from Democracy Works is now live and AP vote counts will appear after polls close in the United States and Brazil. The suppliers are authoritative, but no public deployment audit shows how reliably ChatGPT presents, updates or corrects their data.
The short version
- What happened
- OpenAI's 20 October update says Democracy Works voting guidance is live in ChatGPT across US languages, while Associated Press vote counts will appear after polls close for elections in the United States and Brazil.
- Who it affects
- Voters may get faster, multilingual access to registration and polling-place information, but a confident response can still omit local qualifications. Time-sensitive logistics should be checked against the linked election authority before acting.
- What remains unproven
- All three sources are announcements from parties to the integrations; none independently audits the deployed ChatGPT experience.
Keep in mind
All three sources are announcements from parties to the integrations; none independently audits the deployed ChatGPT experience.
No query volume, tested-language denominator, answer-level accuracy, correction latency or user-outcome measure is published.
AP's 2024 race-call accuracy claim describes AP's process, not the reliability of OpenAI's ingestion, labelling or conversational presentation.
The public material does not show how the interface handles recounts, unresolved races, conflicting location context or emergency changes to voting logistics.
Multi-source analysis · 3 linked sources. Primary analysed source date: 20 October 2026. Earliest linked supporting source: 27 May 2026.
Keep following the evidence
Join The Impact Brief founding list for selected AI reports, their effect on people and the evidence limits. Read a sample first.
Read the full report · 4 min The evidence, methods, implications and limitations
Two established data suppliers are moving inside the chatbot
OpenAI's 20 October update says its Democracy Works integration is now live in the United States across languages. When people ask about registration, deadlines, voting locations, identification rules or other requirements, ChatGPT can draw on information that Democracy Works compiles with state and local election offices. OpenAI also says Associated Press vote counts will appear after polls close for elections in the United States and Brazil. The AP agreement itself was announced on 27 May and covers US national, state and selected local results through the 2028 general election.[1][2][3]
A live feed does not transfer legal authority to ChatGPT
The source roles matter. Election officials remain authoritative for eligibility, procedures and certified outcomes. Democracy Works is an intermediary that monitors official changes and distributes logistics through its Elections API. AP reports tabulated votes and makes race calls under its own process, but an AP call is not a legal certification. ChatGPT is the presentation layer. A clear answer should retain those distinctions, show the original source and avoid turning an estimate, partial count or media projection into a certified result.[1][2][3]
The distinction that matters
Who is responsible for which election claim?
The same answer can contain information from three layers with different authority.
Election officials
Set rules, publish logistics and certify outcomes
Remain the legal authority
Democracy Works and AP
Distribute official logistics or report vote counts
Provide structured source information
ChatGPT
Find and present the supplied information conversationally
Adds access, not independent authority
The deployment evidence stops at the partners' announcements
OpenAI reports that the Democracy Works feature is live, but none of the three organisations publishes how many queries have used it, what languages were tested, how often answers match the underlying record or how quickly corrections propagate. There is no public evaluation covering ambiguous place names, rapidly changed polling sites, close races, recounts or conflicting user context. AP cites 99.9% accuracy across nearly 7,000 race calls in the 2024 general election, but that is AP's historical performance claim—not a measure of the new ChatGPT interface.[1][2][3]
Voters gain convenience, but verification remains one click away
The practical benefit is speed: a person can ask in ordinary language and receive linked information without knowing which local office to search. That may be especially useful for multilingual voters or people navigating unfamiliar rules. The risk is misplaced confidence in a fluent answer. For a voting deadline, location or identification requirement, the safest final check remains the linked state or local election authority. For election-night numbers, readers should look for the reporting source, percentage counted and whether a result is a projection or a certification.[1][3]
What would change the assessment
A useful public audit would sample real election questions across states, municipalities and languages; compare answers with time-stamped official records; measure citation preservation and correction latency; and publish failures as well as averages. Until then, the integrations are a consequential distribution change with credible upstream suppliers, not proof that conversational election information is consistently complete or error-free.[1][2][3]
Watch the background, then check the evidence
Illustrative photo · not a video frame
The Royal Society · 44 min 28 sec · Background explainer
Is this the AI future we want? | The Royal Society
YouTube loads only when you choose; playback starts when you press play.
A Royal Society discussion with Madhumita Murgia and biomedical engineer Alison Noble on AI’s potential in healthcare, science and culture, recorded at the 30 June 2026 event.
Evidence check: A discussion of opportunities and risks, including future possibilities. It is not evidence that every application described is available or clinically proven.
Published 2026-07-01. Source checked 14 October 2026. Check captions on the publisher’s video page.
Check the healthcare evidencePhoto: Daniel Sone / National Cancer Institute / Unsplash · Licence
Continue the story
Read next
Continue exploring elections: compare the finding and its evidence limits.
What are AI chatbots telling voters? MIT has logged 3.75 million answers—but not voter effects
A live MIT audit repeatedly asks election questions across model, search and identity settings. The scale is unusual, but the team has not yet completed its longitudinal analysis or tested how real voters respond.
Read the next reportExplore the full source and evidence record
What this means for people
- Voters may get faster, multilingual access to registration and polling-place information, but a confident response can still omit local qualifications. Time-sensitive logistics should be checked against the linked election authority before acting.
Global context
Election-information chatbots increasingly sit between voters and official records. The design question is not only source quality but whether the interface preserves provenance, uncertainty and changing status across jurisdictions and languages.
What the evidence does not yet show
- All three sources are announcements from parties to the integrations; none independently audits the deployed ChatGPT experience.
- No query volume, tested-language denominator, answer-level accuracy, correction latency or user-outcome measure is published.
- AP's 2024 race-call accuracy claim describes AP's process, not the reliability of OpenAI's ingestion, labelling or conversational presentation.
- The public material does not show how the interface handles recounts, unresolved races, conflicting location context or emergency changes to voting logistics.
What to watch next
- A public multilingual audit against time-stamped election records
- Visible labels separating partial counts, projections and certified results
- Correction-latency reporting when an official source changes
- Published incident and appeal routes for election-information errors
- Evidence on whether voters reach and understand the linked official sources
Living evidence record
Impact record IAI-0A9UQPX
Evidence stage
Announced
Confidence
Supported
Reporting basis
Multi-source analysis
Independent or research support
Not yet
Record status
Monitoring
Last checked
22 October 2026
Source trail
3 direct sources across 1 source type.
People impact
Documented in this record.
Uncertainty
Limits and next checks are explicit.
Stages describe the evidence available—not whether a technology is good or bad. See the public method.
Evidence trail
Sources used for this report
Links checked 22 October 2026
Research question, topics and regions
Topics: Elections · Civic information · Vote counts · ChatGPT · Media provenance
Regions: United States · Brazil · North America · South America
This report is labelled multi-source analysis. We summarise and analyse source material in our own words; company statements remain attributed claims until independently supported. Translated summaries preserve the meaning of the original source and link back to it. Read our editorial standards.
Reader commentary
What does this mean where you are?
Tell us what you have seen at work, in your community or in everyday life. A question is welcome too. No account is needed; we review notes before publishing them.
Explore commentary across the portal →Published reader notes
0No published reader notes yet. You can start the evidence-led discussion above.
Prefer a private correction or response? Contact the newsroom.