Stop Re-Investigating the Same Wallet: A Shared Evidence Layer for On-Chain Crime
Forensics for on-chain crime still runs on manual correlation: you have a wallet, a domain and a file hash, and no way to check whether any of them already appeared in someone else's case.
In five minutes I'll share the one design choice that mattered most when we built a shared evidence layer: keeping raw artefacts separate from published intelligence, so pooling evidence doesn't quietly create a surveillance dataset. Then how AI scoring routes each submission to auto-approve, human review, or reject — and how we defend against reputation-farming.
Takeaway: a reusable pattern for pooling forensic evidence without poisoning it, plus honest numbers on what the AI layer still gets wrong.
Speakers: Dmytro Matviiv
Track: Security
