The OpenExpo Virtual Experience 2021 took place at the beginning of this month as we told you. The virtual event was a success despite the fact that after the pandemic people do not have as much appreciation for virtual events as they did before the pandemic. More of 140 speakers and more than 4000 attendees endorse this success but during the conference We were able to see other successes that are not as noticeable but just as important as attendance.
A good example of this is the lecture he gave Chema Alonso. The famous cybersecurity specialist spoke to us on this occasion about the great danger that represent deepfakes today and in the future.
A deepfake is a modification of a video or an image by which, thanks to AI, it seems that a person is doing something or is in a certain situation. This poses a danger not only to the privacy and honor of people but also represents a danger for programs that use facial recognition as an access method, which in many films is usually a spoofing.
Table of Contents
The DeepFakes has grown considerably in the last year
Chema Alonso warned us that this practice is in full growth and that is dangerous. Until July 2019 the number of DeepFakes circulating on the internet was 15.000, a year later, the number of deepfakes had increased to 50.000 and it is a number that continues to increase. The least bad thing about it is that so far, 96% of deepfakes correspond to pornographic content and focus on influencers and famous people. This is relatively less bad because these people are usually surrounded by someone, have a public life and are easy to deny and detect, but this does not mean that this technology cannot be applied to other situations and / or people.
OpenExpo Virtual Experience 2021 has gathered more than 4.000 participants
As we have said time and again, the best thing about Free Software is the Community that is created around it, and Chema Alonso's exhibition is a great example of this.
Given the fear or damage that a DeepFake can cause, Chema has indicated what methods can be used to detect a DeepFake: detection by forensic analysis of images and extraction of biological data from images. In addition, Chema has indicated that he is working on a plugin for Chromium that I would use these principles so that any user can recognize a DeepFake from their browser.
The plugin for Chromium is still in production but it will use the following tools for proper operation: FaceForensics ++ (a DeepFakes database that will grow as we pass suspicious videos or images); Exposing DeepFake Videos by Detecting Face Warping Artifacts (Since DeepFakes make images with very low resolutions, this tool checks that the image corresponds to the original resolution); Exposing Deep Fakes Using Inconsistent Head Poses (looks for inconsistencies in the 3D modeling and thanks to the HopeNet model a statistical check is made between the different calculated vectors); CNN-generated images are surprisingly easy to spot… for now (the image created between a CNN base bank of images and it is searched if the generated image is related to this base). This will make the Chrome plugin you are working on a great security tool and a tool that will be enhanced by your community.
What we have seen in Chema Alonso's presentation is just a sample of what we can find in the OpenExpo YouTube channel, where we will find the recordings of the talks, events and / or conferences of the event. We can even find the trivia games that were played between the speakers and that they do not stop being entertaining.
Let's hope that next year the event will not only be repeated but also it can be in person and have something online for those of us who cannot be in a face-to-face mode, that is, to take the good from physical events and online events.
To finish I want pick up some words that Chema Alonso mentioned and that they are to reflect on the problems of computer security: “this begins to be real, Black Mirror Style. If we cannot trust what we see, what is left for us?