Deep Fake AI: Its Possible Threats and Tips to Avoid It -

What is Deep Fake AI?

Deep fake is one of form of artificial intelligence that is used to create noax convincing images, videos and sounds. The word deep fake is combination of two words which means the deep learning concept with something fake. This technology is mostly used for notorious purposes such as spreading false propaganda, incorrect information to misguide the public etc.

In this article, you will know that how deep fake AI works, for which purposes it is being used and some tips that how you can spot it from real images or videos.

Uses of Deep Fake AI

Deep fake is being widely used for multiple purposes such as;

Political Manipulation

Deep fake videos are being used to spread the fake videos and news of famous world leaders like Barack Obama and Donald Trump that were concerned with their use in election manipulation. For example, it was concerned that deep fake videos would affect the 2020 election campaign in U.S.

Celebrities Deep Fake Poses

Major threat of deep fake is its use in the non-consensual images and videos of celebrities that make almost 96% of deep fakes on internet.

Hoaxes and Scams

Cyber criminal uses deep fake to create false hoaxes, claims and scams. For example, hacker will make video of senior director about his any criminal activity or make false allegations about company. It can negatively affect he brand reputation, its share price etc.

Identity Frauds and Financial Theft

Deep fake AI can be used to steal the real identities or can be used to create new identity. Deepfake experts use this technology to create false voices of victims or documents to create account by pretending that person.

How You Can Spot Deep Fake AI?

Deep fakes can be recognized by spotting the unnatural movements or unusual activity such as;

Unnatural Body Shape

If a person’s body does not look like having natural body shape, it is likely fake because deep fake mainly focus on face instead of focusing entire body that create unnatural body shape.

Abnormal Skin Color

Deep fake is not able to copy the exact skin tone or color of natural images or videos which leads to abnormal skin tone.

Bad Lip Syncing

Videos created by deep fake AI would not synch properly with the words spoken by the people in video.

Inconsistent Facial Positions

Images created by deep fake will feature awkward-looking or inconsistent body and head positioning. For example, distorted images and jerky movements of people when they turn or move their heads.

Unnatural Hair

Fake images are able to generate realistic features of individuals, such as messed-up or frizzy hair.

Unusual Eye Blinking

A lack of proper eye movement is sign of deep fakes. Unusual eye movement is challenging as eyes of peoples usually react and follow to the person whom they are speaking with.

How You Can Combat with Deep Fake AI?

Several steps have already been taken to help the people from becoming the victim of deep fake AI. Some of them are following;

Research Lab Technologies

IT researchers are working on data science solutions that detect deep fake AI. Many of them have become ineffective quickly as the technology of attackers also evolves and creates more convincing results.

Rules of Social Media

Facebook has hired several IT researchers from different universities to build a deep fake detector, which can help to ban deep fakes. Twitter has lunched new policies in order to prevent fake content and is working to detect deep fake videos and images that are not immediately removed. YouTube also worked to block deep fake content related to the U.S. 2020 elections and census.

United Sates Legislations

Several U.S. states have already been passed laws to criminalize deep fake and prevent its use around elections. Deep fake AI legislation was also introduced in December 2019 into the National Defense Authorization Act (NDAA).

Filtering Programs

Filtering programs are also working to prevent use of deep fake AI. AI firm Deep Trace’s program acts in the same way as spam or antivirus filter and diverts the fake content into a quarantine zone, while another AI Foundation; Reality Defender has an aim to tag manipulated content before it can do any damage.