ClaimReview Schema for AI Trust: Specification and Implementation

TL;DR

Apply ClaimReview to any article that asserts a verifiable claim. Required fields: claimReviewed, datePublished, author, reviewedBy, itemReviewed. AI engines weight ClaimReview strongly in regulated and high-stakes verticals.

Background

ClaimReview was introduced by schema.org for fact-checking publishers (Snopes, PolitiFact). Google honored it in classical search via the "Fact Check" rich result. With the rise of AI search, the schema's value broadened: any verifiable assertion gains trust signal lift when wrapped in ClaimReview.

Specification

{
  "@context": "https://schema.org",
  "@type": "ClaimReview",
  "datePublished": "2026-04-28",
  "url": "https://example.com/article#claim-1",
  "claimReviewed": "BSA Section 314(b) permits voluntary information sharing among covered financial institutions.",
  "itemReviewed": {
    "@type": "Claim",
    "datePublished": "2026-04-28",
    "appearance": "https://example.com/article",
    "author": {
      "@type": "Person",
      "name": "Jane Doe",
      "sameAs": ["https://www.linkedin.com/in/janedoe"]
    }
  },
  "author": {
    "@type": "Person",
    "name": "Jane Doe",
    "sameAs": ["https://www.linkedin.com/in/janedoe"]
  },
  "reviewedBy": {
    "@type": "Person",
    "name": "John Smith",
    "jobTitle": "Chief Compliance Officer",
    "sameAs": ["https://www.linkedin.com/in/johnsmith"]
  }
}

Field requirements

Patterns

Pattern A: Single-claim article

One ClaimReview per article, applied to the article's main thesis. Best for definitions, regulatory explainers, single-fact references.

Pattern B: Multiple claims with anchor URLs

Multiple ClaimReview blocks, each with its own #anchor URL. Best for FAQ pages, multi-claim guides, technical references.

Pattern C: Authoritative Statement

Non-fact-check publishers using ClaimReview to assert their own authoritative position. The reviewer is an in-house subject-matter expert with credentials.

Validation

• Use Google's Rich Results Test.
• Validate JSON-LD shape with schema.org validator.
• Confirm reviewer credentials are verifiable through sameAs links.
• Check that the claim text is the same as a sentence in the article body.

Anti-patterns

• Vague claims ("This is a great solution.") — not verifiable.
• Marketing puffery wrapped in ClaimReview — erodes trust if engines detect.
• Reviewer = author — reduces independence signal; either separate them or omit reviewedBy.
• Missing sameAs on author/reviewer — fails entity disambiguation.
• Mismatched datePublished between root and itemReviewed.

When ClaimReview is not appropriate

• Subjective claims (opinion, taste).
• Fictional content.
• Marketing copy.
• Bare lists without verifiable assertions.

How to apply

1. Identify articles that make verifiable, high-stakes claims.
2. Extract the canonical claim sentence.
3. Add ClaimReview JSON-LD with all required fields.
4. Add a named reviewer with credentials and sameAs.
5. Validate and ship; monitor citation lift over 60 days.

FAQ

Q: Will Google Rich Results show ClaimReview to non-fact-check publishers?

Generally no — the rich result UI is restricted to recognized fact-checkers. The AI search trust signal is independent and does benefit non-fact-check publishers.

Q: Can I add ClaimReview to existing articles?

Yes. Updating older articles with ClaimReview gives a freshness + trust lift; pair with dateModified update.

Q: Does ClaimReview hurt SEO?

No — invalid implementations may be ignored, but valid implementations carry no penalty.

Q: Can the reviewer be an AI?

No. reviewedBy should be a real person or organization with verifiable credentials.

Q: How many ClaimReview blocks per page?

Up to ~5 typically. More than that risks structural noise; split into multiple pages instead.