amazon rekognition biometrics congress

Deepfake and the privacy

Deepfake is a technology which based on a machine learning technique called GAN(Generative Adversarial Network).Using this technique, anyone can  produce or alter video contents or audio recordings that look and sound just like real things.

However, this Deepfake has been mostly used for applying celebrities faces on porn stars body and also making funny stories using  politician’s and popular people’s speeches.

But the terrible thing is  , using this technique , it is very straight forward to create fake emergency alert warnings or generating fake audio and video that can revers the results of close elections or break someones’s marriage with fake sex videos.


How it works

Deepfake is based on GAN(Generative Adversarial Network) a kind of machine learning technique.A GAN is  created by combining two AI systems.One is ‘generator’ and other is ‘discriminator’.

Firstly creator should train this system till it gets enough accuracy.Generator generates a fake content and pass to discriminator to verify whether the content is fake or real.Then discriminator identifies that the content is fake and provide feedback to generator to adjust  it parameters to improve the accuracy.

Above process occurs in both training and actual using scenarios. This helps system to improve it accuracy with the increase of it usage.


The real threat

Deepfake can meltdown someones’s privacy one way or another. Although this system needs bunch of photos to generate more realistic content,it is easier to find those photos by examine anyone’s social media feeds.

Today also social media crimes as at it highest.So can you imagine what would happen when these AI tools are more generalized.That’s why people should identify this as existential treat for democracy and humanity.


Leave a Reply

Your email address will not be published. Required fields are marked *