From the State Government edition of October 27, 2026
States United study: ChatGPT and Google AI cut factual errors on voter questions to 0% in testing, but pointed voters to state election websites in under half of answers and often gave incomplete candidate lists
States United Democracy Center · Election information integrity; evaluation of AI platform answers for state election officials · United States (Arizona, Michigan, Nevada, North Carolina, Pennsylvania, Wisconsin tested)
- Publisher
- States United Democracy Center
- Original publication
- 2026-08-04
- Source retrieved
- 2026-10-28
What happened
States United tested ChatGPT and Google AI on common voter questions in two rounds: 497 responses across six states in Sept-Oct 2025 and 402 responses across Arizona, Michigan and Pennsylvania in Jan-Feb 2026. Verifiable factual error rates fell from 6.9% (Google AI) and 8.2% (ChatGPT) to 0% in the second round, but answers were often incomplete. ChatGPT mentioned a state election site in 39.4% of responses and Google AI in 55.6%. ChatGPT returned incomplete candidate lists for 88.9% of gubernatorial queries. Of 3,481 cited links, state and local government sources made up 41.5% and Wikipedia 12.3%. Google AI switched to links-only answers around Feb. 2, 2026. Recommendations for election officials include structuring official sites for AI discoverability and building voter education around candidate filing windows.
Why it matters
Why now: published in August but newly relevant because Ohio's secretary of state launched an AI error-reporting channel on Oct 27, a week before the Nov. 3 election. It is the independent evidence check on that approach: platforms have become more accurate on simple facts, and the bigger gap is that they often do not send voters to official state sources. Not in our archive.
Evidence and measured results
Full report page opened, including findings, recommendations and methodology: standardized questions coded by researchers across two rounds and three platform conditions. Limits: two platforms, a few states, testing before the general-election period; not peer reviewed.
Limitations and uncertainty
Advocacy research organization; small state sample; platform behavior changes quickly, as the Google format change shows. Results may not hold for Claude, Grok or other platforms named by Ohio.
Put this evidence to work
Lighthouse Advisory interpretation, grounded in this source. Enriched 2026-10-27; this does not change the original publication date. Each role follows the same checklist so takeaways compare across sources.
Sales
Role takeaway
- Customer problem
- Voters use AI platforms for election questions, but those platforms send voters to official state sites less than half the time and give incomplete candidate information.
- Stakeholders
- Secretary of state or state election director
- Election office web and communications team
- State CIO digital services team
- County election officials
- Discovery questions
- Does the official election site use structured data and stable URLs?
- Are candidate filings published in machine-readable form?
- Has the office tested how AI platforms answer its voters' questions?
- Who would own outreach to platforms?
- Value hypothesis
- Interpretation: better-structured official content could raise the share of AI answers citing the state site; the report offers this as a recommendation, not a tested result.
- Potential engagement
- Bounded AI discoverability assessment of one state's election website with a repeatable platform test using standardized voter questions.
- Unsupported claims
- AI platforms are not shown to be fully reliable
- Site changes are not proven to raise citation rates
- Results may not apply to platforms not tested
Pre-sales engineering
Role takeaway
- Fit
- State election websites and candidate filing systems that AI platforms draw on for voter answers.
- Architecture and integration
- Structured data markup on key pages; machine-readable candidate filing and election date feeds; consistent URL conventions; scheduled automated query tests against major AI platforms with citation logging.
- Prerequisites
- Content inventory of voter information pages
- Candidate filing data access
- Standardized question set per state
- Staff to review test results
- Constraints
- No control over platform behavior
- Platform formats change without notice
- Candidate fields change during filing windows
- Security
- Limited relevance; ensure official feeds are authenticated and tamper-resistant so platforms cite accurate data.
- Proof of value
- Measure the share of AI answers citing the state election site before and after content changes
- Measure candidate list completeness during a filing window
- Track error and incompleteness rates across repeated test rounds
Delivery
Role takeaway
- Work
- Improve official election content structure, publish machine-readable data, run periodic AI platform tests and align voter education with the election calendar.
- Dependencies
- Web team capacity
- Candidate filing system data access
- Election calendar
- Platform outreach contacts
- Ownership
- State election office communications and web team, with the state digital services or CIO team for site changes.
- Skills and adoption
- Web staff need skills in structured data and AI search behavior; communications staff need a routine for testing platform answers.
- Governance checkpoints
- Baseline platform test before changes
- Retest at filing deadlines and before certification
- Post-cycle review
- Acceptance criteria
- Proposed: AI answer citation rate to the state site measured each cycle
- Proposed: candidate filing data available in machine-readable form during filing windows
- Proposed: standardized test repeated at least twice per cycle
- Risks
- Platforms change formats and undo gains
- Effort focuses on accuracy while incompleteness persists
- Staff capacity during election season
Implementation considerations
Lighthouse Advisory interpretation across the operating dimensions a public-sector buyer must settle before this evidence becomes a design. Each note answers the question under its heading for this specific source.
Architecture and integration
What must connect, and where does the AI sit in the workflow?
For state election offices, the practical lever is content structure: structured data markup, plain language and stable URLs on official sites so AI platforms retrieve and cite them, as the report recommends.
Governance
Who approves, reviews and stays accountable for outcomes?
The report frames AI as a complement to official information and puts most fixes on platform operators (always link the state election site, prefer official sources, integrate candidate filing data); states can only influence these through outreach.
Security and privacy
What data, permissions and controls need testing?
Limited relevance; the account-level personalization test (election-skeptic browsing profile) concerns information integrity rather than data security.
Accessibility and workforce
Who is affected, and what skills or accommodations follow?
Recommends plain-language official content; adds periodic testing work for election web and communications staff.
Procurement
What should contracts, pricing and exit terms secure?
Not addressed.
Operating model
Which teams own the service once it runs?
Positions state election offices as publishers optimizing for AI retrieval and as testers of platform answers, alongside traditional voter education.
Publication history
- 2026-10-27State Government · Issue 503 resources
Stable resource ID: states-united-ai-platforms-direct-voters-to-state-election-sites-under-half-20260804