Colombia. Digital fraud continues to grow, surpassing the ability of state and financial institutions to react. In this scenario, artificial intelligence (AI) is beginning to consolidate itself as an essential resource for predicting threats, identifying irregularities and minimizing both operational and fiscal risks.
One of the most complicated scenarios is that of artificial identities, generated by mixing authentic data with invented information. These profiles, as they have no direct relationship with the victims, are more complicated to identify and can be maintained for long periods without being recognized.
The figures reflect the increase in this threat. In the first half of 2024, 6.9% of digital transactions in the country were considered suspicious of fraud, representing an increase of 43.5% compared to the same period last year, according to data from TransUnion. Colombia was among the five countries with the highest rate of attempted digital fraud in the world, while four out of ten citizens said they had been the target of scams through banking, social or commercial channels.
The pressure on public finances is also increasing. Although tax evasion has historically represented a challenge for the Colombian State – with estimates that exceed 50 billion pesos per year, according to the National University – today it is intensified due to technological practices that are difficult to follow, such as the misuse of subsidies, false documents or digital fraud.
Against this backdrop, experts agree that anticipation should replace the reactive approach. "Today fraud operates in real time, with automated tools. To face it, institutions need technology that detects, decides and acts in advance. And that can only be achieved with advanced analytics and artificial intelligence applied with purpose," said Ricardo Saponara, leader of risk, fraud and compliance advisory for Latin America at SAS.
Technologies being implemented include:
Artificial data, which makes it easy to create synthetic replicas of authentic data to train predictive models without compromising sensitive information.
Decision-making automation, capable of handling large amounts of information in milliseconds and triggering alerts against unusual operations.
However, the adoption of these technologies requires a solid ethical foundation. "Analytics must be governable, traceable and fair. The decisions a system makes must be able to be explained. It's not just about efficiency, it's about public trust, protecting the most vulnerable, and tax justice," Saponara warned.
A global study by Economist Impact in partnership with SAS showed that 80% of executives in the financial sector expect financial crime to have a severe operational impact in the next decade. While 99% already implement generative AI solutions, more than 50% have yet to achieve concrete financial benefits.
