Multimodal
Deepfake
2017ActiveUpdated: 15 September 2026Published
Key
innovation
Using generative models to create realistic synthetic media impersonating real people or events.
Category
Multimodal
Abstraction level
Pattern
Operation level
InferenceApplication
Use cases
Special effects and dubbingDisinformation and fraud (misuse)Development of media detectors and watermarking
How it works
A model learns the features of a person's face, voice or body from data, then generates new frames or audio, matching facial expressions, lip movement and vocal timbre. In the GAN approach a generator competes with a discriminator; diffusion models gradually denoise toward a realistic output. Increasingly, multimodal models generate coherent audio and video.
Problem solved
A deepfake is itself a generative capability; as a phenomenon it drives the development of detection, watermarking and media authenticity verification methods.
Evolution
2017
The term deepfake emerges
Inflection pointPopularization of neural-network-based face-swapping techniques.