Introduction:
Every morning we wake up to new technology or invention that takes the tech
world a step or even multiple steps forward, and every day we seem to be
confronted with something new. Recently, we've seen the emergence of deepfake
technology which has completely upended the way we think about media, as well
as the way, we deal with news and verify the validity of the information we absorb
online.
world a step or even multiple steps forward, and every day we seem to be
confronted with something new. Recently, we've seen the emergence of deepfake
technology which has completely upended the way we think about media, as well
as the way, we deal with news and verify the validity of the information we absorb
online.
It's nearly impossible to look at the world in the same way we did just a few years
ago, as we all have to now deal with these new truths that we never expected to
exist, Deepfakes are a prime example of this, especially in an era of such fast-paced
tech, the advancement of the deep fake is being developed at an alarming rate.
So what this technology really is, and how dangerous it can be if used the wrong
way?!
What's the Deep Fake?
The deep fake is a machine learning technology that can create artificial videos of
anyone saying anything they want it to say.
anyone saying anything they want it to say.
It's dangerously advanced as the generated videos and images can't be
differentiated from real ones, in addition to the availability of the real-time deep fake
technology which allows the person to replace their face with any other face,
including the face of a celebrity, politicians, and public figures the thing that creates
the possibility to easily manipulate media & public opinion and even make the
the opportunity of carrying out real destructive actions really high.
By using a series of neural networks and techniques that can extract facial
movements from audio or video, deep fakes can use the power of Al to generate
pre-recorded and real-time videos that can be hardly exposed.
But how exactly can a deepfake be created?
A deep fake can be created using two different technologies: Motion capture and
facial keying.
facial keying.
With the motion capture technique, two video streams are created: An input feed
which is a recording of a person performing a particular action and a projection
feed (of the same action) which is a recording of a person performing the same
action but from another angle. The projection feed is then processed to create a
"fake" video of the input feed, with the fake video containing the same movements
as the real video.
The second technique involves matching the faces in a video to those of another
person. This is achieved by using a facial recognition algorithm to identify key
points on the face of the person in the video and track the movement of the face as
it moves between different frames in the video. The face is then digitally altered to
reposition the facial features, such as the position of the eyes, nose, and mouth.
These repositioned features are then matched to the facial data recorded from the
other person's video and the gaps filled between the frames to produce a
new "fake" video of the person speaking in it.
"Deepfake" technology was pioneered by researchers at Stanford University in
2015.
What's The DeepFake Offensive Toolkit (DOT)?
The deep fake toolkit is an open-source toolkit developed using python and a bunch
of other software packages such as OpenCV, and it allows you to play around and
create fake videos using the DeepFake technology.
of other software packages such as OpenCV, and it allows you to play around and
create fake videos using the DeepFake technology.
It allows the user to change the details of the destination live (Real Time) for videos
or even live broadcasts. Using Virtual Camera Injection. They were built mainly as a
kind of Reverse engineering or reverse engineering to detect and verify
videos that use Deep Fake technologies in a biometric form.
The tool takes the content broadcasted through your camera, and then it concocts
the Deep Fake content conveniently and live (Real Time) so that even when you
move in front of the camera the system does not fail to make a fake video of the
image of your face. Then through the Virtual Camera Injection fake content is
broadcast to others.
the Deep Fake content conveniently and live (Real Time) so that even when you
move in front of the camera the system does not fail to make a fake video of the
image of your face. Then through the Virtual Camera Injection fake content is
broadcast to others.
The DOT system does not need any kind of training or provide intelligence with a
a lot of content at all, all it needs is a photo, then take a video through the camera,
And it will directly do face swapping & change efficiently.
Is it possible to try DOT myself?
The simple answer is, yes! of course, you and anyone can install the toolkit and play
around with it as it's open-source and available on Github.
around with it as it's open-source and available on Github.
The toolkit is available on both Linux and Windows, tho configuring it can be a little
bit tricky and requires some understanding, thus it's recommended to check the in-
detail explanation on how to configure the tool depending on the system you have.
Take into account that you will need to install python, the project, as well as a bunch
of other projects. As for Windows, the OBS Studio software is needed as it helps
with the use of the camera with the VirtualCam option.
For further instructions on how to install and configure the tool, check out the DOT-
Project/GitHub.
Project/GitHub.
Finally, once you have all set, you're ready to begin the journey of discovering the
power of the tool and mess around with videos with the help of the Al!
Conclusion:
The DeepFake isn't a simple or limited technology, and as visible it's going to keep
evolving and making progress, as well as raising lots of questions as to ethical, legal,
and manipulation aspects. Leaving us with the big question that is floating on the
surface more and more every day, do we still have cyber privacy as internet users
anymore?!
evolving and making progress, as well as raising lots of questions as to ethical, legal,
and manipulation aspects. Leaving us with the big question that is floating on the
surface more and more every day, do we still have cyber privacy as internet users
anymore?!