AI at the service of scammers: how not to become a victim of neural networks

Since the emergence of publicly available generative AI models in 2022, they have been massively implemented in all sectors of human activity. The trend towards neural networks has not gone unnoticed by scammers, who have significantly expanded their toolkit thanks to the new technology. GetBlock AML Research explains for what purposes attackers use AI, and how to protect your assets from thieves in these conditions.

On both sides

Since its advent, AI technology has been used on both sides of the barricades. Neural networks are not only helping scammers but also the experts who counter them. Law enforcement agencies and analytics companies are also incorporating AI into their work to increase efficiency. Neural network-based tools can significantly increase the speed and efficiency of data processing. Neural networks are being used to create advanced detection systems that can scan the blockchain in real-time for suspicious and illegal transactions.

Popular uses of AI by scammers

• Deepfakes. The attackers create fake audio recordings and videos with famous public figures, in which they urge them to invest in dubious projects. The most common deepfakes are those with Elon Musk, Donald Trump, and other media personalities. The danger of such deepfakes is that they are virtually indistinguishable from real videos, so many users may believe the information spread by deepfakes;
• AI for phishing. Neural networks are massively used to create fake identities and elaborate disguises. With the help of AI, scammers create highly plausible emails, fake websites, and chatbots that convince potential victims to perform dangerous actions;
• KYC bypass. Using neural networks, attackers can create fake document scans to verify identities on major trading platforms;
• Real-time. AI allows scammers to spoof voices and faces in real-time audio or video calls. This was a tactic used by North Korean hackers Lazarus Group to get interviews at large financial firms.

How experts disclose the use of AI by scammers

In addition to standard research methods, analysts now use the following techniques:

• Deep pattern recognition, which can be used to determine the authenticity of a video recording;
• Linguistic analysis recognizes audio tracks and identifies atypical artifacts, such as unusual pronunciation of words and strange accents;
• Stylistic features: experts compare a person’s speech style with a deepfake and identify inconsistencies.

All this data combined with transaction chains in the blockchain can give a complete picture of how scammers operate, what tools they use, and what results they achieve.

Comparison of Elon Musk’s original video with the deepfake

How to protect yourself from AI used in criminal activities

The methods listed below will help you identify potentially dangerous offers from scammers, who often rely on the quick reaction of a potential victim. This is why attackers neglect the quality of deepfakes.

• In AI-generated videos, pay attention to unnatural facial expressions, blinks, and inconsistencies in lip movement to the pronunciation of words;
• Lighting and shadows should also be examined, as neural networks often mistakenly place shadow artifacts that do not match the lighting;
• When receiving AI-generated phishing emails, links, and domain names should be carefully checked, as they contain intentional alterations that direct the victim to malicious websites;
• Financial firms should train their employees in basic skills to counter social engineering.