Financial fraud in Colombia: human manipulation before money moves
Financial fraud is evolving towards the manipulation of human decision-making, before any transaction occurs. Bre-B surpasses 1.7 billion transactions and faces new threats from social engineering.
How financial fraud acts before a single peso moves
A call that appears to come from the bank, a message asking to confirm a transfer, or a voice imitating someone trusted: these are the tools of modern financial fraud. According to semana.com, fraud no longer depends primarily on exploiting technological vulnerabilities, but on manipulating human decisions. In many cases, the transfer is not the start of the attack, but its last visible step.
The deception begins when someone manages to impersonate an identity, alter the context, or induce a decision. In an environment where money can move in seconds, anticipating such situations has become one of the main challenges for banking.
Bre-B: a year of instant payments and new risks
Colombia has concrete reasons to pay attention to this shift. Bre-B, the Central Bank's instant payment system, completed its first year with more than 36 million registered users and more than 1.7 billion transactions processed. The advance deepens financial inclusion and connects millions of people and businesses, but it also creates new protection demands in each interaction.
Fraud methods already exploit citizens' familiarity with digital payments. During 2026, Colombian authorities warned about text messages impersonating Bre-B requesting users to click fraudulent links to confirm alleged transfers, with the aim of obtaining sensitive information. The risk can begin long before an operation reaches a financial institution's validation systems.
Deepfakes and artificial intelligence serving deception
Added to this is a broader transformation. According to the Anti-Fraud Technology Benchmarking Report 2026, produced by SAS and the Association of Certified Fraud Examiners (ACFE), 77% of fraud prevention professionals surveyed reported an increase in social engineering methods using deepfakes over the past two years. However, only 7% believed their organisation was more than moderately prepared to detect or prevent frauds driven by artificial intelligence.
The challenge goes beyond making deception more convincing. Artificial intelligence can also be used to impersonate identities, forge documents, or attempt to breach authentication mechanisms. Tools that transform productivity can equally make fraud strategies more personalised and difficult to identify.
Four directions for banking security
Facing this reality, banking security must evolve in four directions, according to semana.com's analysis.
Understand behaviour, not just verify the transaction. Controls based on amounts, locations, times and frequency remain necessary, but advanced analytics allows for incorporating additional signals: unusual changes in a client's habits, unknown devices, new recipients, or suspicious relationships between accounts. The SAS and ACFE report indicates that 54% of respondents already combine internal and external data, but only 34% use unstructured internal data. This shows room to expand analytical signals and shift the question from "does this transfer appear fraudulent?" to "is the behaviour leading to this transfer consistent with this client?".
A high-value transfer is not necessarily fraudulent, just as a small transaction can be part of a deception pattern. Artificial intelligence and machine learning models allow for identifying anomalies, analysing relationships and assigning risk levels with greater precision.
Bring prevention to the same speed as payments. In instant payment systems, detecting a suspicious operation when money has already left severely limits intervention possibilities. Real-time analytics allows for evaluating risk signals during the operation and determining when additional validation is necessary or when to trigger an alert.
Protecting should not mean making the user experience difficult. A strategy that blocks transactions indiscriminately generates friction, affects trust and limits the benefits of digitalisation. The challenge lies in combining predictive models, business rules and knowledge of client behaviour to strengthen security without turning each payment into a complex process.
Build a shared vision of risk. People use banks, digital wallets, e-commerce and different channels to interact with the financial system. Criminals can exploit that diversity to distribute their activities across platforms and entities. Prevention requires better integrating available information, strengthening collaboration mechanisms between ecosystem actors and developing models capable of recognising patterns that are not always evident when analysing a transaction in isolation. The same SAS and ACFE study shows that such collaboration can significantly expand the field of observation.
Adapt controls to the pace of fraud. The fourth direction, implicit in the analysis, suggests that financial institutions continuously update their detection tools as defrauders' tactics evolve, without waiting for damage to become visible in account statements.
Source: Google News CO — Crime