A deepfake is digitally generated or altered media that convincingly depicts a person, event, audio recording, or video that is not authentic. Deepfakes use deep learning to fabricate synthetic media, creating opportunities for filmmaking while enabling deception and manipulation. section 1.3 section 10.2
Deepfakes create practical risks when audiences trust electronic media without adequate verification:
- Influence elections by portraying political opponents in compromising situations. section 10.2
- Manipulate markets through a fabricated corporate announcement, such as a false bankruptcy statement from a chief executive. section 10.2
- Intensify cyberbullying by creating convincing abusive or humiliating media about individuals. section 1.3 section 10.2
A fabricated bankruptcy announcement
A neural network trained on Elon Musk’s interviews could generate a persuasive video making him appear to announce Tesla’s bankruptcy. Before the deception was exposed, it could cause severe financial harm, illicit profit, or public chaos. section 10.2
The central danger is not only that deepfakes can look or sound authentic, but that increasingly electronic communication makes deceptive media easier to distribute and harder to contain after publication. section 10.2
