Deepfake: The Most Famous Direction in AI Fraud.
Deepfakes are perhaps the most famous and widely discussed tool that fraudsters use with the help of artificial intelligence. These technologies allow the creation of convincing fake videos and audio that are difficult to distinguish from real ones.
The good news: If your photos or videos rarely appear online, you are unlikely to be targeted. To create a high-quality deepfake, scammers need a significant amount of material:
​
-
Optimal amount: At least 30–60 minutes of video showing you speaking, moving, and expressing various emotions.
-
For audio: 1–2 hours of diverse voice recordings.
​
Without such data, the likelihood of creating a realistic fake drops significantly.
How Could Deepfakes Affect You?
​
While deepfakes often feature public figures, they can also target ordinary people. Imagine receiving a video call from someone who looks and sounds exactly like your child, urgently requesting money or sensitive information. Scammers use AI to create these realistic impersonations, making it challenging to distinguish between genuine and fake communications. This technology enables fraudsters to exploit your trust and emotions, leading to potential financial loss or compromised personal data.
What is a Deepfake?
​
A deepfake is a synthetic video or audio clip created using artificial intelligence that convincingly mimics real people. At the core of this technology are Generative Adversarial Networks (GAN) — a type of AI that improves itself through competition. GAN consists of two parts:
-
Generator: Creates fake data, such as a face image.
-
Discriminator: Evaluates how realistic the generated data is.
These two models work as rivals: the generator learns to fool the discriminator by producing increasingly realistic fakes, while the discriminator gets better at spotting them. This iterative process allows GAN to create content that is nearly indistinguishable from real data.
How Do Deepfakes Work?
​
Deepfakes are created using GAN and involve the following steps:
-
Data Collection: Gathering large amounts of audio and video samples of the target individual.
-
Training the AI: Feeding this data into the GAN so that the generator learns voice patterns, facial expressions, and mannerisms.
-
Content Creation: The generator produces new, realistic footage or audio of the person performing actions or speaking words they never actually did.
For example, in 2018, a well-known deepfake video featured former U.S. President Barack Obama delivering a statement he never made. This video was created by filmmaker Jordan Peele to highlight the potential dangers of deepfake technology.
Why Are Deepfakes Dangerous?
​
The risks associated with deepfakes include:
-
Misinformation: Spreading false information that can influence public opinion or incite unrest.
-
Fraud: Impersonating individuals to deceive others into financial transactions or revealing confidential information.
-
Reputation Damage: Creating fabricated content to tarnish someone's personal or professional image.
As deepfake technology becomes more sophisticated, it is increasingly important to remain vigilant and verify the authenticity of unexpected or suspicious communications.