“Look here, I have the rehearsals, there is a video.” Can we still say in the era of the deepfakes generated by the AI? Of course, any recovery, even earlier, could be false or manipulated, but not to this point and this speed.
Already on social media it is difficult to understand if the video with a strange animal is true or not (so much so that they have put the “obligation” to indicate whether a video is generated or not with the AI), as well as you do not It can use for example chatgpt to generate images with real characters, not even with yours (I wanted an image of me in the arms of Freddie Mercury, or the opposite, I tried to convince it, always replies that, for privacy, cannot generate images with existing people).
Grok generates you, very similar, but by changing some connotations on purpose. For example, if you ask him to generate, who I know, an image of Trump who sings with a helmet of bananas on his head as Freddie Mercury (a random example) makes it, but observing well you realize that it is not really Trump, it is one species of impersonator. From the results of a survey by the Connected Consumer Study of Deloitte it shows that 59% of the interviewees could not understand the difference of a content generated by the human being and one with the AI.
These, however, for now, returning to images and videos of people, the rules of engagement of the AI of common use, but at other levels you can do everything. According to the Entrust Cybersecurity Institute in a very short time we would have a deepfake attack every five minutes.
That’s why we are thinking of countermeasures, and who can be put against the AI if not the same Ai, another to? In April Honor will launch his function to Deepfake Detection. He will examine the Pixel images and videos for pixels, imperfection by imperfection, the consistency of each individual movement, notifying users of videos and materials potentially manipulated by the AI, which should discourage Deepfake seeders (Gigi Ballarani, during a conference on Ai, however, said: “Not even the death penalty has ever discouraged criminals”, and in fact).
However, I see it hard, it will be like the long struggle between virus and antivirus, because you know, with technology and the web cannot be controlled everything.
An anti-deepfake detectioni will be created, and another anti-anti-deepfake detection, and so on. For the moment, that I know, if your wife or husband catch a photo or video in which you are with the lover, you can still say, shutting up, “is a deepfake”. Until April. Then we will see.