$150,000 in fraud loss from one rented contributor account, stopped in three weeks

PLAY THE FILM · 1:44
For trust and safety teams at AI data labeling platforms

in fraud loss from one rented account Stopped in three weeks

SCROLL
The problem

Somebody qualified for expert work, built a real track record, then rented the account to a stranger, and the platform kept paying expert rates

The case · 1 of 7

It passed screening

A real applicant clears the platform's screening on the first attempt, exactly as designed

The pattern

It is never just one account

Every case we have worked reduces to one of six behaviours, and they all end in the same place: the platform pays somebody other than the person it approved

  1. 01

    Account resale

    Somebody passes your screening, then sells or rents the account to a second person

    Verification layer only
  2. 02

    Multi-accounting

    One person runs several contributor accounts at once, so you pay one person six times

    Verification layer only
  3. 03

    Location concealment

    The account claims a country you approve and pay rates for while the work happens elsewhere

    Platform infrastructure
  4. 04

    Outsourced verification

    The account holder forwards your verification link and asks somebody else to complete it

    Both, if joined
  5. 05

    Check avoidance

    The contributor reaches the check, understands what it asks for, and stops

    Verification layer only
  6. 06

    Spoofing

    A photograph, a screen, or a partly obscured face is presented to the camera

    Verification layer only

Four of the six leave traces that only a verification layer can see

The fix

One person, one account

HumanCheck confirms that a contributor is a real, unique, live person, and compares every check against every check that has run before it

  • Account per person
    1

    Every check is compared against every check that ran before it

  • Seconds, start to finish
    3-5

    A live check in the browser through one API call

  • Documents
    None

    No government ID and no uploads. VerifyYou does not store images

  • Ranked accounts acted on
    4 of 4

    One offboarded and three formally warned by the platform

Next step

Read what we found, or see it on your own contributor pool

White paper cover
White paper · 13 pages · 14 minute read

Get the white paper

  • The six behaviours and the trace each one leaves
  • Why the four controls most platforms run cannot see them
  • The signal hiding in your verification drop-off
  • How verifying at the door changes the economics
Talk to the team

Book a demo

If you run a contributor pool and want to know what a verification round would surface in it, this is the shortest path

  • One API call at application, first payout, or periodic re-check
  • You keep your own screening, tests and reviewers
  • You get a ranked short list with the evidence attached
The product

HumanCheck, explained by our CEO

David Graham, CEO and co-founder of VerifyYou, walks through what HumanCheck does and where it fits in your product

  • “VerifyYou helps prevent AI fraud and abuse, with much lower friction for end users”
    Wes MichaelPresident and Founder, Rare Patient Voice, now part of Konovo
  • “There needs to be something in this space to convince researchers like me that what you got is from a person”
    Gary BallingerProfessor of Commerce, UVA McIntire School of Commerce
  • “Instead of forcing the disclosure of AI, verify and mark the human”
    Dr Alan FinkelFormer Chief Scientist of Australia, Founder of Proudly Human

Find out how many of your contributors are the same person

Tell us about your contributor pool and we’ll run a first verification round with you and walk through what comes back