Verify a decision

{"actor":"judge","decided_at":"2026-08-25T17:04:34.915Z","decision":"review_revise","investigation_id":"6a80ffa9-764c-4d6c-819b-2534f22d1540","new_score":0,"page_slug":"july-2026-bridge-hack-wave-three-protocols-35-6m-one-day","prev_score":0,"reason":"Using the reviewer's authoritative claim_findings array (32 entries: 23 confirmed, 5 partially_supported, 4 disputed, 0 unverifiable), disputed_pct is 12.5%, placing this page in the low end of the minor-issues band. The core per-incident facts — dollar amounts, dates, attack mechanics, and fund flows for AFX Trade, Verus, and B-Squared — are well-corroborated across multiple independent outlets, and the headline $35.6M figure does arithmetically reconcile with the three per-incident totals (claim_findings[1]). The page also correctly avoids blaming the underlying chains (Arbitrum, Ethereum, BNB) for what were bridge- or contract-specific failures, and its dedicated 'coordination' section properly discloses that no security firm has linked the three incidents to a single actor. However, three distinct sourcing defects push the page above a clean approval: it attributes the catchy 'Hackers Day' label specifically to Blockaid when no source supports that specific attribution (claim_findings[4]); it asserts, in both a narrative section and the timeline, that Verus knowingly redeposited ~$11.58M into an unpatched bridge on July 8 — a claim its own cited source does not contain and that is otherwise sourced only to an uncited opinion blog (claim_findings[13], [26]); and it states H1 2026 set records for both incident count and aggregate dollar losses, when its own cited sources (Blockaid, TRM Labs) explicitly say losses were lower than the prior year and only the incident count was a record (claim_findings[3]). Separately, the page's own title and entity_name ('Bridge Hack Wave — Three Protocols') mischaracterize B-Squared, which independent reporting describes as a staking-contract exploit rather than a bridge hack (claim_findings[32], flagged as a high-priority coverage gap) — a defect in the page's core framing that isn't captured in the disputed count. Together these issues do not undermine the page's substance but they materially affect the story's framing (an invented 'known reckless redeposit' narrative about Verus, an inflated H1 record claim, and a mislabeled title), warranting revision rather than approval.","score_delta":-14,"sequence_num":3,"submission_content_hash":null,"submission_id":null,"submission_kind":null,"submission_valence":null,"v":1}