AI changes scam warning signs; Unico and Liminal estimate 6 million victims in 2025

The advance of artificial intelligence is making scams more accessible and making it harder to identify fraud through visual clues. A report by Unico and Liminal estimates 6 million account-theft victims in 2025, up 141% from 2021; the injection of manipulated images and videos represents 45.8% of classified cases. Tools for sophisticated attacks can cost about US$30 a month, according to g1. São Paulo accounts for 22.06% of fraud alerts in Brazil.

Rodolfo Avelino, an Insper professor interviewed by g1, says flaws such as deformed fingers and a lack of blinking are becoming less reliable. He recommends ending the conversation, calling a previously known number and asking a question only the real person would know how to answer. Unexpected requests for money, emotional urgency, pressure to act and attempts to prevent confirmation through another channel are more important warning signs. Luis Felipe Monteiro, Unico’s CEO for Latin America, says AI has lowered the barriers to carrying out these crimes.