Criminals use facial comparison AI to commit identity fraud in Brazil
Criminals in Brazil reportedly use facial comparison AI tools to locate people who physically resemble them, then attempt to impersonate those individuals during identity verification to commit fraud. Serasa Experian identified this tactic, called the "lookalike scam," and reported blocking 2.86 million such attempts in early 2026, preventing an estimated R$ 3.77 billion in losses.
What happened
Criminals are using facial comparison technology to locate people with physical characteristics similar to their own, then attempting to impersonate those individuals during identity validation processes. The strategy, termed 'golpe do sósia' (lookalike scam), exploits facial recognition systems by having fraudsters pass themselves off as other people during account openings and registrations.
According to Serasa Experian's analytical intelligence team, criminals use tools capable of comparing their own face against images in various databases, including illegally obtained ones, to identify legitimate identities with high physical similarity. The fraudster then selects an identity deemed most likely to pass biometric validation and attempts to use that person's data in fraud schemes.
Reported impact
Serasa Experian's anti-fraud solutions identified and blocked 2.86 million identity fraud attempts in the first half of 2026, a 32.9% increase compared to the same period in 2025. The blocked attempts were associated with an estimated R$ 3.77 billion in prevented losses during the period.
Index analysis
The new tactic reverses traditional identity fraud logic: criminals begin with their own face, search for physically similar people, identify potential victims, then attempt to use that victim's identity data in fraud. According to Leandro Bartolassi, director of Authentication and Fraud Prevention at Serasa Experian, the strategy demonstrates how criminals can exploit technologies designed to increase security.
Governance controls
Serasa Experian recommends that facial recognition not be used as the sole authentication layer. The company advocates combining facial recognition with other signals including document validation, liveness checks, data verification, device analysis, behavioral pattern identification, and additional risk signal evaluation to detect inconsistencies that facial analysis alone would miss.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.