Deep Fakes are mainly manipulated videos in which the face of a person acting in a video is replaced by the face of another person, the manipulated video looks deceptively real. For example, one takes the recorded video interview of a politician and replaces the face of the politician with the face of another person. Deep fakes also include videos in which the acting persons makes statements she has never really made (with perfect lip sync). Deep fakes are generated at the help of neural networks with high computational effort.

It is also possible to generate audio tracks that genuinely simulate a person's voice, based on a surprisingly small voice samples.

The potential for abuse of this technology is obvious. For example, the faces of prominent women are projected into pornographic videos. There is also a case where a finance employee made a money transfer to a third-party account after receiving a supposed call from a board member with the instructions to do so; the caller had genuinely faked the board's voice as a deep fake.


The author is a manager in the software industry with international expertise: Authorized officer at one of the large consulting firms - Responsible for setting up an IT development center at the Bangalore offshore location - Director M&A at a software company in Berlin.