É real? A nova era das fraudes por IA.

When Fraud Looks Like a Movie Scene

For years, realistic face manipulation belonged to films like Mission Impossible or Face/Off. What once required Hollywood budgets now costs less than $6 per month using Fraud-as-a-Service tools powered by generative AI. Today, the same techniques are being weaponized by fraudsters.

‍

The Numbers Behind the Threat

Recent data reveals a dramatic acceleration in advanced fraud attempts.

Unico Research (2024-2025). Data based on proprietary analysis of AI-accelerated fraud attempts and market projections.

Global Fraud Trends:

• +1082% Increase in sophisticated fraud attempts in 2025

• +150% Expected growth in 2026 under conservative projections

• Up to +550% Growth in AI-accelerated scenarios

What we are witnessing is not just more fraud — it is smarter fraud. The whitepaper analyzes real attack data and explains why generative AI is rapidly lowering the barrier to entry for cybercriminals.

‍

Anatomy of a Modern Identity Attack

Modern biometric fraud typically falls into two major categories:

Presentation Attacks (Physical spoofing)

Fraudsters attempt to deceive the device camera using physical artifacts.

Common techniques include:

• Deepfake videos displayed on curved monitors to simulate 3D depth
• Optical manipulation with mirrors to distort reflections and bypass texture analysis
• High-resolution screen playback designed to mimic live behavior

Injection Attacks (Digital bypass)

Instead of tricking the camera, attackers bypass it completely: In 2025: 45.5% of sophisticated attacks occurred through injection techniques.

Examples include:

• FaceSwap: Overlaying a victim’s facial features onto the attacker.
• Blending: Merging real human traits with victim biometrics to produce a hybrid identity capable of passing weaker verification systems.

These attacks are invisible to traditional camera-based liveness detection.

‍

The Technology Powering the New Fraud Economy