Verify a decision

{"actor":"judge","decided_at":"2026-08-25T22:18:45.528Z","decision":"review_revise","investigation_id":"396643eb-ec66-4db4-a8c0-189283b3d26c","new_score":0,"page_slug":"neyro-token-aurum-foundation-exit-scam-rebrand","prev_score":2,"reason":"The reviewer checked 41 atomic claims and found zero disputed by contradicting sources; the mechanical disputed_pct (disputed + unverifiable / total) is 3/41 = 7.3%, which sits inside the approve band, and the page's central allegations — that Aurum Foundation was a Ponzi scheme that rebranded to Neyro and then exit-scammed investors on July 30, 2026 — are directly and repeatedly confirmed by tier-2 sources via direct fetch (claim_findings[2], [7], [10], [40]). However, two things the disputed_pct formula cannot see push this into revise rather than approve. First, claim_findings[18] shows the page states as settled fact that Shane Morand's prior company was 'connected to a $13.7 million Vietnamese Ponzi scheme,' when the reviewer found the only supporting BehindMLM article poses that link as a question in its own headline; on a page that names several living individuals in connection with fraud, presenting a hedged allegation as established fact is a real defect even though it is scored 'partially_supported' rather than 'disputed.' Second, coverage_gaps[0] (rated high priority) notes that this page and the separate 'aurum-foundation' page cover the same operation with no cross-reference between them, which risks giving readers an incomplete or inconsistent picture of the same entity's trust score — a structural gap, not a factual error, but one the review rules direct toward a revise outcome even at low disputed_pct. Secondary issues also warrant correction but are lower-stakes: an operator quote in claim_findings[38] is paraphrased rather than verbatim ('system update' vs. the source's 'technical maintenance'), two timeline dates (claim_findings[24], [25]) carry invented day-level precision not supported by the cited source, one regulatory date (claim_findings[29]) cites the wrong article, and the Isaacs LinkedIn-removal claim (claim_findings[9]) rests solely on an unverified reader comment that a follow-up comment in the same thread contradicts.","score_delta":-8,"sequence_num":3,"submission_content_hash":null,"submission_id":null,"submission_kind":null,"submission_valence":null,"v":1}