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
- Sequence
- #2
- Score
- 2 → 2 (0)
- Cluster
- mainnet-beta
- Slot
- 443512087
- Off-chain at
- 2026-08-25T07:13:55.790Z
- Anchored at
- —
- Block time
- —
Independent verification
- 1. Database (off-chain)
- 6vHYscaumaiuk4vSmNBzbADVZHrJLeAxotQHrbEDbGvn
- 2. Recomputed (your browser)
- computing…
- 3. On-chain (Solana memo)
- fetching…
Canonical bytes hashed (1949 chars)
{"actor":"reviewer","decided_at":"2026-08-25T07:13:55.631Z","decision":"review","investigation_id":"e458fdb1-9e2d-42a4-9b2a-a13e95f69e1d","new_score":2,"page_slug":"ai-powered-crypto-phishing-infrastructure-2026","prev_score":2,"reason":"Of 52 atomic claims checked, 34 were confirmed and 8 were partially supported by their cited sources. 6 claims were disputed — including a CertiK phishing-loss growth figure contradicted by CertiK's own H1 2026 report, a Group-IB/FBI '312%' AI-fraud figure that does not match the widely-reported 1,210% figure from the same underlying data, an ACM paper misquoted on drainer-affiliate commissions (paper says 10-30%, page says ~80%), a Hong Kong voice-cloning case with a currency/figure error (HK$18.5M vs. the actual HK$145M/~US$18.5M), a deepfake-detection-rate claim attached to a source (IEEE Spectrum) that does not discuss the topic at all, and a timeline date error placing a January 2026 Chainalysis report a year early. 4 claims — including a precise 52%/$1.13B deepfake-impersonation statistic, a Chainalysis '$341 million single wallet' figure, and Operation Atlantic's '30+ countries' scope — could not be traced to any cited or independently locatable source. Several other figures (e.g. the $682,000 vs. $719,000 non-AI scam revenue baseline, and a March 2026 OFAC/North Korea citation pointing to an unrelated TRM Labs article) reflect minor sourcing/citation errors rather than fabrication. The large majority of the page's core statistics (Chainalysis's $17B/4.5x/253%/1,400% figures, the Netcraft GitBook and Lighthouse/Lucid PhaaS findings, FBI IC3 AI-fraud figures, Operation Atlantic's headline numbers, OFAC Cambodia sanctions, FraudGPT/GhostGPT/WormGPT details, and the OpenAI Cambodia scam-network disruptions) were independently corroborated.","score_delta":0,"sequence_num":2,"submission_content_hash":null,"submission_id":null,"submission_kind":null,"submission_valence":null,"v":1}