Doctored videos or deepfakes have been one of the key weapons used in propaganda battles for quite some time now. Donald Trump taunting Belgium for remaining in the Paris climate agreement, David Beckham speaking fluently in nine languages, Mao Zedong singing 'I will survive' or Jeff Bezos and Elon Musk in a pilot episode of Star Trek... all these videos have gone viral despite being fake, or because they were deepfakes.
 |
| Guess what is his real age and nationality answer is in the photo of the blonde girl |
India's envoy to UN, TS Tirumurti said, "Terrorist and radical extremist groups today have unprecedented access to the general public through the internet, which allows for more efficient and effective recruitment, incitement, and propaganda, as well as the purchase of weapons and illegal money transfers."
India has warned the world at United Nations Security Council (UNSC) that artificial intelligence-enabled 'deep learning' to create 'deepfakes' is being used for misinformation across the World and terror groups are using the internet for many things from recruitment to purchase of weapons
The potential danger of deepfakes lies in the fact that the level of manipulation is so perfect that it can be seemingly impossible at times to distinguish them from real videos. And the more difficult it becomes to detect the falsity, the greater the threat it possesses to pass off as real and cause the havoc it intends to. But with more sophisticated tools powered by artificial intelligence available now to produce these videos, is it becoming more difficult to detect deepfakes?
What are deepfakes and how are they created?
Deepfakes constitute fake content - often in the form of videos but also other media formats such as pictures or audio-created using powerful artificial intelligence tools. They are called deepfakes because they use deep learning technology, a branch of machine learning that applies neural net simulation to massive data sets, to create fake content.
It is hard to make a good deepfake on a standard computer. Most are created on high-end desktops with powerful graphics cards or better still with computing power in the cloud. This reduces the processing time from days and weeks to hours. It also takes expertise to touch up completed videos and to reduce flicker and other visual defects.
 |
| Faces of 3 individual make a new face with the help of deepfake |
It employs a branch of artificial intelligence where if a computer is fed enough data, it can generate fakes which behave much like a real person. For instance, Al can learn what a source face looks like and then transpose it onto another target to perform a face swap.
The application of a technology called Generative Adversarial Networks (GAN), which uses two Al algorithms-where one generates the fake content and the other grades its efforts, teaching the system to be better-has helped come up with more accurate deepfakes.
GAN can also come up with computer-generated images of fake human beings, which has been used by a website called "This Person Does Not Exist. This makes it virtually impossible to detect whether the videos or images we see on the Internet are real or fake
Where are Deepfakes Prevalent?
According to AI firm Deeptrace appx 15,000 deepfake videos are posted online every month. 96% of them are pornographic in nature and in 99% of those deepfakes are mapped faces of female celebrities on to porn stars.
Will deepfakes wreak havoc?
We can expect more deepfakes that harass, intimidate, demean, undermine and destabilise. However, India has warned of Deepfakes as bigger problem. India envoy said, "Both state and non-state actors" can use "artificial intelligence-enabled deep learning to create 'deepfakes, which have the potential to fuel misinformation, divisions, and political instability." Also, "advances in AI and 3D printing" could facilitate attacks by "automating the development and deployment of weapons and weapon systems".
With a proliferation of deepfake videos, there is a growing concern that they will be weaponised to run political campaigns and can be exploited by authoritarian regimes.
Last year, before the Delhi Assembly polls, videos of Delhi BJP president Manoj Tiwari speaking in English and Haryanvi went viral. In these videos, Tiwari was seen criticising Arvind Kejriwal and asking people to vote for BJP. The videos, which were shared in over 5,000 WhatsApp groups, were later revealed to be deepfake.
Deepfakes are also a cause for concern at a time when WHO has stated that the Covid-19 crisis has triggered an infodemic and there have been "deliberate attempts to disseminate wrong information to undermine the public health response and advance alternative agendas of groups or individuals".
 |
| The pic of that man and this girl is created through deepfake in reality these two people do not exist |
Moreover, doctored videos - which includes manipulating the content by using incorrect date stamp or location, clipping content to change the context, omission, splicing and fabrication are increasingly used nowadays on social media to deliberately misrepresent facts for political ends. Most of these videos are not examples of deepfakes but show how easy it can be to obfuscate facts and spread lies based on manipulated content masquerading as hard evidence.
The threat posed by Deepfake videos is already apparent.
"There are malicious users using such videos to defame famous personalities, spread disinformation, influence elections and polarise people. With more convincing and accessible deepfake video synthesis techniques, this threat has become even bigger in magnitude,"
Most social media companies such as Facebook and Twitter have banned deepfake videos. They have said as soon as they detect any video as a deepfake, it will be taken down.
Facebook has recruited researchers from Berkeley, Oxford, and other institutions to build a deepfake detector. In 2019, it held a Deepfake Detection Challenge partnering with industry leaders and academic experts during which a unique dataset consisting of more than 100,000 videos was created and shared.
However, not all deepfakes can be detected accurately and it can also take considerable time for them to be found and taken down.
How do you spot a deepfake?
It gets harder as the technology improves..US researchers discovered that deepfake faces don't blink normally. Poor-quality deepfakes are easier to spot,The lip synching might be bad, or the skin tone patchy. Flickering around the edges of transposed faces, Fine details, such as hair, Jewellery etc are particularly hard for deepfakes to render as well. Recently, the first Deepfake Detection Challenge was conducted by Microsoft, Facebook and Amazon.
0 Comments