Verify a decision
Every moderation decision on AVOID.NET is anchored to the Solana blockchain. You don't have to trust us — you can verify cryptographically that we committed to a verdict at a specific moment and have not rewritten it.
How verification works
- We commit. When a moderator accepts/rejects a submission, we serialize the decision into deterministic UTF-8 bytes (
payload_canonical_string), hash it with SHA-256, encode the digest as base58, and write it to Solana inside an SPL Memo v2 transaction. - We store the bytes. The exact bytes we hashed are stored alongside the decision in our database. Anyone can read them and recompute the hash in any language.
- You compare three values. Database hash, your independently-recomputed hash, and the hash inside the on-chain memo. If all three match, the decision is authentic and timestamped.
The on-chain memo format is
AVOID.NET|v1|h:<b58-sha256>|d:<id>|t:<iso>Find a signature on any investigation page's decision log, or run python -m src.verify_decision --signature <sig> for a CLI check.
Decision
review_revise · AI-Powered Crypto Phishing Infrastructure 2026
- Sequence
- #3
- Score
- 2 → 0 (-12)
- Cluster
- mainnet-beta
- Slot
- 443512091
- Off-chain at
- 2026-08-25T07:13:55.890Z
- Anchored at
- —
- Block time
- —
Independent verification
- 1. Database (off-chain)
- JA4t9kxWWMUCeWM1RmeDqbLESugLELPKRYFWnJJ6W1d8
- 2. Recomputed (your browser)
- computing…
- 3. On-chain (Solana memo)
- fetching…
Canonical bytes hashed (2315 chars)
{"actor":"judge","decided_at":"2026-08-25T07:13:55.631Z","decision":"review_revise","investigation_id":"e458fdb1-9e2d-42a4-9b2a-a13e95f69e1d","new_score":0,"page_slug":"ai-powered-crypto-phishing-infrastructure-2026","prev_score":2,"reason":"Of 52 claims checked, 34 were confirmed and 8 partially supported, but 6 were disputed (19.2%) and a further 4 were unverifiable. The disputed claims are not confined to trivial details: claim_findings[4] has CertiK's own H1 2026 report showing phishing losses down about 11% year-over-year, directly contradicting the page's '~200% increase' claim; claim_findings[11] misstates a widely reported 1,210% FBI/IC3 AI-fraud surge as '312%'; claim_findings[34] states an 80% drainer-affiliate commission where the cited academic paper documents 10-30%; and claim_findings[48] attaches a deepfake-detection-rate claim to an IEEE Spectrum article that does not discuss the topic. claim_findings[12] crosses a currency figure (stating HK$18.5 million where the real case was HK$145 million, equivalent to roughly US$18.5 million), and claim_findings[49] places a January 2026 report a year early on the timeline. Because this page exists mainly to convey the scale of AI-enabled fraud, errors in headline magnitude and percentage figures sit closer to the page's core purpose than a typical peripheral detail would, and the reviewer separately flagged a high-priority coverage gap about the page's reliance on single, uncorroborated vendor statistics for exactly these numbers. Weighed against this, reviewer confidence was the lowest in this review batch (0.65), several fetch/access failures limited direct verification, and the large majority of the page's other statistics -- including its most load-bearing figures ($17B in 2025 scam losses, the 4.5x AI premium, Operation Atlantic's disruption totals, IC3's AI-fraud figures, and the Lighthouse/Lucid phishing-as-a-service findings) -- were independently corroborated. This combination supports a revision requirement rather than denial: the page's overall narrative holds up, but a cluster of specific statistics needs correction before the page can be considered fully reliable.","score_delta":-12,"sequence_num":3,"submission_content_hash":null,"submission_id":null,"submission_kind":null,"submission_valence":null,"v":1}