The other side of Artificial Intelligence: seven financial fraud threats that use this technology are identified

The proliferation and accessibility of Artificial Intelligence tools promoted a new stage of accelerated development of digital financial services. But at the same time it is being used by cybercriminals to innovate your fraud tactics too at unprecedented speed.

This technological development goes hand in hand with new social engineering strategiesa technique used by those who commit crimes to manipulate people and get them to voluntarily disclose their personal data and bank passwords, or transfer money to fake contacts.

He 98% of cyber crimes involves some type of social engineering, according to data collected by the company Lynx Tec. The permanent renewal of deceptions through social engineering forces the continuous updating of anti-fraud solutions to keep up with new threats, and those that do not. they risk becoming ineffective.

One of the main challenges for financial institutions, whether fintech or traditional bankswill be to ensure a balance between adopting fraud prevention measures and offering a customer-friendly experience, according to experts. For this, they must implement systems capable of recognize habitual purchasing patternsavoiding the need for unnecessary additional authentications.

Furthermore, given the certainty that cybercriminals are diversifying their fraud targets to other markets and activities, The need to carry out analysis and prevention of financial crimes will grow in sectors beyond banks, fintechs and payment methods companies.

Artificial intelligence at the service of cybercrime

In Latin America, cybercriminals use various advanced techniques to carry out financial fraud. A survey of the aforementioned company specialized in these topics identified the seven most common strategies that threaten the financial sector and also to other online activities that handle large volumes of operations.

Phishing and Spear Phishing: With the help of AI, cybercriminals create emails and messages that imitate legitimate communications, personalizing them to increase their effectiveness. They also use voices generated by AI (vishing) to impersonate executives, celebrities or officials, seeking to extract sensitive information or make transfers.

Advertising fraud (Ad Fraud): AI-powered bots generate false clicks and impressions on online advertisements, causing losses for advertisers. These bots imitate human behavior, making it difficult to detect them and generating erroneous perceptions about non-existent products or services.

E-commerce fraud: Cybercriminals create fake online stores or clone legitimate ones, tricking customers into making fraudulent purchases. Automated systems are also used to generate fake reviews and ratings, manipulating product rankings and reputations.

Synthetic identity fraud: With different tools on the Internet, criminals create false identities by combining real data available on social networks and fictitious ones, which they use to open bank accounts, apply for loans or commit other types of financial fraud. This technique is difficult to detect, as synthetic identities can include real information, making it difficult to trace illicit activities.

Deepfakes: The manipulation of videos and audios using AI, known as deepfakes, allows criminals to impersonate key people to carry out fraud or extortion. In more recent cases, deepfakes are used to execute social engineering attacks, convincing victims to make money transfers or share sensitive information.

Payments in real time: The growing demand for instant payments has allowed criminals to take advantage of this speed to receive funds immediately, making it almost impossible to recover money from victims by deceiving them with false payment receipts. Furthermore, this speed facilitates money laundering, dispersing funds across multiple accounts in a matter of minutes, making it extremely difficult to track and putting users at risk.

Account Takeover Attacks– Cybercriminals use AI to execute account takeover attacks, leveraging databases of leaked credentials or information collected through phishing. Once an account is compromised, they can make unauthorized transactions or steal personal information.

“To confront these threats, it is essential that banks and financial institutions adopt AI and machine learning in their anti-fraud systemsconstantly updating its detection models. They must also encourage collaboration between actors in the financial ecosystembreaking information silos to improve global visibility,” explained Juan Pablo Jiménez Isaza, vice president of Lynx Tech for Latin America.

What are the cybercrime trends for next year?

At the same time, the Lynx Tech study also detected some of the main trends in fraud, technologies for analysis and risk mitigation that will gain strength in 2025:

Instant payments: Fraud associated with the growth of real-time payment platforms, such as Pix in Brazil, CoDi in Mexico, and Bre-B in Colombia, will increase significantly. These transactions, due to their speed, make it difficult to detect and block fraudulent operations or operations associated with false identities. The adoption of advanced solutions based on AI and ML will be essential to identify suspicious patterns and prevent attacks in real time, adapting to the evolution of threats in markets with regulations still developing.

Centralized approach: A clear trend will be observed towards the centralization of multiple transactional channels, in the same model for detecting and preventing fraudulent movements, creating a unified vision of the customer journey. This approach will allow for more effective analysis of behavior and facilitate the detection of inconsistencies, strengthening both security and trust in the financial institution.

Model-as-a-Service: The offer of analysis and detection models as a service to third parties is emerging as one of the big trends for next year. This approach allows companies to access advanced solutions without the need to develop or maintain technological infrastructure internally.