Inside the Arms Race to Stop AI Deepfakes From Beating Identity Checks

A long vendor search

Emiliano "Nano" Rodriguez says Bitso, which has operated in the region for twelve years, built "one of the strongest, maybe longest RFPs" in the industry and evaluated more than 30 vendors across the region. He names Incode's strong LatAm footprint as one of the main differentiators against the others.

From probabilistic to deterministic

Iñigo Castillo says Incode pioneered verification against government databases in Mexico. Checking an applicant against a governmental source of truth, he says, turns identification from a probabilistic process based on algorithms into a deterministic one. He says Incode is also the first company able to connect to several US state databases.

Castillo also argues that owning its full stack sets Incode apart from competitors that bundle third-party technology, because "fraud is changing so fast. You need to adapt really fast."

How the fraud works

Iñigo Castillo describes fraudsters who get a jailbroken phone, install a virtual camera and slide deepfake videos into the onboarding process. He says Incode was built ten years ago to verify identity fully automatically, expecting a time when the human eye could no longer do it. He also cites a recent credit bureau leak in one of Incode's markets that let fraudsters request credit in other people's names. He adds that biometric re-authentication, such as a device change, opens a window for account takeover.

Preparing for what comes next

Emiliano "Nano" Rodriguez predicts that as AI evolves at an exponential rate, fraud will become very hard to detect on digital platforms. He advises peers to be prepared and to work with a robust identity partner. Iñigo Castillo expects the next frontier to be agents: with people running hundreds of agents that move funds, he says, processes will be needed to identify who is behind them.