How to Protect Against Deepfake Attacks

Nov 30, 2021
3 min read

Imagine having a video call with someone known to you as a relative and giving that person personal and confidential information about yourself but getting to know in the end that the facial image you thought was known to you was fake.

The development of AI(artificial intelligence) in the tech world has brought about a different form of developmental penetration. Almost all sectors make use of artificial intelligence in carrying out their day-to-day activities. Hence, this has led to a high proliferation in the development and advancement in the use of Artificial Intelligence. With this, a different quagmire has emerged from a form of artificial intelligence called Deep Fake.

Deep flake is a deceptive piece of content videos and audio recordings that has been manipulated or fashioned through artificial AI. Deepfakes are very compelling— and successfully trap people into a full conviction that a person did or said something that never happened. It uses the factual facial, voice and the overall convincing appearance of a personality to commit illicit purposes habitually.

Although deep fake technology is not very tricky to the world and not really complicated enough to deceive the public at scale. However, some very believable clips of prominent individuals like Barrack Obama, Queen Elizabeth, Mark Zuckerberg etcetera, have made known to people how great and realistic it can be. Deep fake is posing many threats to the world and needs a dire step into protecting against it. Below are the steps to protect against deep fakes.

Learning How to Spot a Deep Fake

Learning how to spot a deep fake gets harder as technology improves; nonetheless, learning how to detect a deep fake is still pertinent. In 2018, US researchers discovered that deep fakes faces do not blink normally. At first, it seemed like a magical solution for the detection problem. But no sooner had the research been published that deep fakes appeared with blinking than how the game works. Poor-quality deep fakes are easier to spot—  the skin tone patchy, or the lip-synching might be bad. There can be details such as hair and flickering around the edges of transposed faces, which are particularly hard for deep fakes to render well, especially where strands are visible on the fringe.

Applying Generative Adversarial Networks

A basic understanding of how deep fakes are created has dire repercussions on every aspect of daily activities. The major establishment of deep fakes is based on understanding a specific branch of artificial intelligence(AI) known as Generative Adversarial Networks.

Currently, the most pertinent aspect of AI is focused on neural networks—a specific type of algorithm that learns in layers to build complex concepts out of less complex ones. Generative Adversarial Networks use two neural networks: one to generate fake images of a person and a second to then evaluate those images. The first neural network (the generator) attempts to create images that deceive the second neural network into thinking they are factual. Also, a neural network is asked to discern if the generated images, mixed in with real images from previously existing data, are authentic or fake. The generator and discriminator vary, with the generator aiming to improve in each cycle until it can usually hoodwink the discriminator.

Investing in New Technologies

Several steps can be applied to reduce the impact of deep fakes, even once they are indistinguishable from real video. A reverse image search process has empowered people to expose original photos from which forgeries are made. The users can upload an image and discover similar photos online using computer vision, revealing the photo as altered using this tool.

Creating Reforms to Ban Deep Fake Contents

.  Creating reforms to ban deep fake content will make it easier for private citizens to legally challenge technology platforms accountable for disseminating slanderous or harmful content that they upload. These qualifications would provide individuals leverage to appeal to technology companies to remove deep faked content of their resemblance from these websites.

Finally, as technology improves, the fight against deep fake gets harder. However, there is of utmost importance in collaborating to fight against this malicious way of representing people in a fake/illegal way. This will create a trusted space for the users of AI and ensure a more safe and secured environment for people and industries that are making use of it.

References

https://www.google.com/amp/s/www.creativebloq.com/amp/features/deepfake-examples
https://www.google.com/amp/s/www.brookings.edu/research/fighting-deepfakes-when-detection-fails/%3famp
https://www.google.com/amp/s/spectrum.ieee.org/amp/what-is-deepfake-2650279986
https://www.theguardian.com/technology/2020/jan/13/what-are-deepfakes-and-how-can-you-spot-them
https://www.google.com/amp/s/searchsecurity.techtarget.com/post/Best-practices-to-detect-and-mitigate-deepfake-attacks%3famp=1
https://threatpost.com/deepfake-attacks-surge-experts-warn/165798
https://www.tessian.com/blog/what-are-deepfakes
https://securityintelligence.com/articles/how-protect-against-deepfake-attacks-extortion/