Claims

GET /api/v1/claims returns Brand FactCheck's verified claims: one row per factual statement an AI answer made about a tracked brand, checked against the workspace's approved facts. Each row carries a run_id joinable to runs, so a claim can always be traced back to the answer that made it.

Filters

Standard parameters (limit, offset, format=csv) work as described in Pagination, filtering, and errors.

Key fields

• claim_id: stable identifier
• run_id, brand_id, engine_id: the answer, brand, and engine the claim came from
• claim_text: the factual statement as extracted from the answer
• verdict: accurate, inaccurate, or unverifiable against the workspace's approved facts
• error_class: for inaccurate claims, hallucination, outdated, or incorrect; null otherwise
• confidence: verification confidence between 0 and 1
• theme_id: the claim theme the statement clusters under, or null
• review_state: auto for trusted automatic verdicts, needs_review while a low-confidence verdict waits for a human, confirmed or corrected after review, suppressed when feedback removed the claim from scoring
• created_at: when the claim was recorded

Rows are ordered by created_at descending with claim_id as the tiebreak, so offset pagination is stable.

Claim themes

GET /api/v1/claim-themes returns the lookup table behind theme_id: the canonical topics claims cluster under (for example "pricing" phrased many ways). It accepts one filter, status (active, or merged for themes folded into another), and exposes theme_id, canonical_label, and status.

Example

curl -H "Authorization: Bearer $AEO_API_KEY" \
  "https://app.citlyze.com/api/v1/claims?verdict=inaccurate&since=2026-08-01"

Use this resource to audit exactly what AI engines state about your brand, which statements are wrong and why, and which themes the errors concentrate in. Window-level accuracy rollups are in accuracy aggregates.