Fraud
How Invitebase makes an AI fraud decision on every referral, what you see for each one, and how to unblock a flagged referral.
The Fraud surface is where you read Invitebase's fraud decisions. Every referral gets an AI fraud decision on the pass that validates it, before money moves. You don't need to approve referrals one by one, but you can review any flagged referral and unblock it.
How referrals resolve
A referral that passes the AI fraud check is approved and moves to payout after any hold window. One that fails is flagged (status rejected): it is not paid or billed unless you review it and unblock it.
What you see
You see the decision and a short, plain-English reason. The score is not shown to you.
Invitebase runs and tunes the AI fraud check, so there is nothing for you to configure.
You still decide what counts as a referral through your qualification gates and what rewards you pay through your campaign configuration.
Reviewing and unblocking
AI fraud decisions are not final. A flagged referral still shows in the end user's referral history, marked as flagged. If you decide it's genuine, unblock it and it pays out normally. You can also unblock every flagged referral for one end user at once. Invitebase keeps the original AI decision and records who unblocked the referral and when.
Rewards and billing
Referrals flagged by the AI fraud check are not validated, are not billed, and their rewards are not paid while they stay flagged. If you unblock a flagged referral, it goes down the normal payout path and is billed like any other validated referral. See Fraud prevention for how automatic screening fits into the referral flow, and Payouts for the payout hold window and approval flow.
Analytics
The fraud chart shows fraud decisions over time and the reward spend stopped before it was paid.
Test-mode referrals never bill or move money. See Environments and test mode.