Artificial Intelligence, Cyber Security, Deepfake, Digital, Generative Adversarial Networks (GAN), Innovation, machine learning, News, Threat Model, Zero Trust
by Irfan

Deepfake – Seeing is no longer believing


What is Deepfake

Deepfakes are fake videos or audio recordings that look and sound just like the real thing, today anyone can download deepfake software and create convincing fake videos in their spare time. Deepfakes exploit this human tendency using generative adversarial networks (GANs), in which two machine learning (ML) models duke it out. One ML model trains on a data set and then creates video forgeries, while the other attempts to detect the forgeries. The forger creates fakes until the other ML model can’t detect the forgery. The larger the set of training data, the easier it is for the forger to create a believable deepfake. 

How to spot a deepfake

  • A lack of blinking Many older deepfake methods failed to mimic the rate at which a person blinks – a problem recent programs have fixed.
  • Face wobble Shimmer or distortion is a giveaway. Also, look for abnormal movements from fixed objects in the frame – a microphone stand or a lamp, for example.
  • Strange behaviour An individual doing something implausible or out of character should always be a red flag.
  • But obvious fakes may not be what they seem It is easy to sow doubt about real footage by adding an inconsistency.
Photograph: Egor Zakharov/YouTube

Is it alarming and cause serious impact?

Ofcourse, it’s serious and below are few evidences

  • Nancy Patricia Pelosi is an American politician serving as speaker of the United States House of Representatives since January 2019. With her fake speech video on one particularly popular website, Politics Watchdog, the video received 2 million views and 45,000 shares. This video didn’t require an AI program. The creator just altered the speed of Pelosi’s speech and raised the pitch of her voice to disguise the manipulation.
  • Deepfake of Mark Zuckerberg apparently talking about how he has “total control of billions of people’s stolen data” and how he “owe[s] it all to Spectre”, the product of a team of satirical artists, went viral as well
  • Oscar-winning director Jordan Peele and his brother-in-law, BuzzFeed CEO Jonah Peretti, created a deepfake of Barack Obama apparently calling Trump a “complete and utter dipshit” to warn of the risks to public discourse.
  • Researchers at Samsung’s AI lab in Moscow published “footage” of Marilyn Monroe, Salvador Dalí and the Mona Lisa, each clip generated from one still image. While it is still fairly easy to discern a deepfake from genuine footage, foolproof fabrications appear to be disconcertingly close

How does it impact cyber security

Deepfake technology is now being used to create high-fidelity phishing attacks where the phishing target (financial institution, healthcare provider, auction site, email provider) is indistinguishable from the real entity. Organizations may need to implement new identity protection measures. Additionally, from an ethical perspective, these onboarding systems should be configured to not only capture the various employee biometrics, but also determine how best to safeguard the information so that it can’t be used to create potential fakes. Longer-term, enterprises will need to explore other options such as use of ledgering, certification and checksums. These will necessitate complex processes and tooling like those used for electronic signing of legal documents.