OFAC red flags
The #1 OFAC Red Flag for AI Agents: Paying a Sanctioned Wallet
An AI agent that executes a payment to an OFAC-sanctioned address has, in most readings, already committed a strict-liability violation. This is the red flag that ends companies. Here is what it looks like and how to make it impossible.
| Red flag | Why it triggers OFAC scrutiny | Risk level |
|---|---|---|
| Direct transfer to a sanctioned address | The agent sends USDC, ETH, or BTC to a wallet on the SDN list. Under OFAC strict liability the transfer itself is the violation — intent is not required. | Critical |
| Agent pays a wallet you previously approved | A counterparty's wallet is added to the SDN list after you whitelisted it. Without a real-time screen on every payment, the agent keeps paying a now-sanctioned address. | Critical |
| Paying an address derived from a sanctioned one | OFAC's 50 Percent Rule can extend to wallets controlled by a sanctioned person even when not individually listed. A clean-looking address can still be a violation. | High |
The control: every red flag above is caught by pre-transaction OFAC screening.
SanctionsAI checks the wallet, name, or jurisdiction against the live SDN list in under 100ms,
before the payment is signed. There is no pattern so clever that it bypasses an address check.
What to do if you see one of these
- Stop the transaction. Do not let the agent retry around the screen.
- Log the event with timestamp, subject, and SDN list version (the audit trail is your defense).
- If a payment already executed, preserve evidence and assess voluntary self-disclosure — it can reduce a penalty by up to 50%.
- Review the agent's control path: was the screen on the actual execution path, or only on the happy path?
Block every red flag before the payment signs
Pre-transaction OFAC screening in under 100ms. Free tier: 5 checks/day, no signup.
Screen a wallet free → See pricing