Deep fake
A deepfake is synthetic or manipulated media, often generated using machine-learning methods, that convincingly depicts people, events, or statements in a way that is not authentic.
Introduction to Generative AI, Second Edition
A deepfake is synthetic or manipulated media generated with machine-learning techniques to depict people, events, or statements inaccurately or deceptively.
- How it is used
- Frames deepfakes as synthetic media used for creative production, privacy protection, deception, scams, and political manipulation.
- How deep it goes
- Defines the category and explains social effects, but stops short of technical generation methods or detection procedures.
- Examples
- Uses fabricated political videos, manipulated celebrity media, election robocalls, cloned-voice scams, and documentary face anonymization.
- What you can do after
- Evaluate deepfakes as creative tools and sources of deception, privacy harm, weakened evidence, and institutional distrust.
Read how Introduction to Generative AI, Second Edition explains deep fake →
Sutskever's List
A deepfake is realistic AI-generated synthetic media depicting people doing or saying things that did not occur.
- How it is used
- Frames deep fakes as an analogy for misinformation concerns surrounding GPT-2’s generated text.
- How deep it goes
- Uses the analogy without explaining how synthetic audiovisual media are made.
- Examples
- Illustrates the concern with headline-generated news articles and the fictional scientist in GPT-2’s “Ovid’s Unicorn” sample.
- What you can do after
- Evaluate/understand why plausible AI-generated text prompted misinformation concerns and affected GPT-2’s release.
Build a Text-to-Image Generator (from Scratch)
A deepfake is synthetic or manipulated media generated or altered with machine-learning methods to impersonate or misrepresent real people or objects.
- How it is used
- Uses deep learning and computer vision to distinguish real images from AI-generated deepfakes.
- How deep it goes
- Explains definition, risks, dataset preparation, transfer learning, final-layer retraining, and unseen-image testing.
- Examples
- Uses real-versus-fake image datasets, unseen test images, and a fine-tuned ResNet50 classifier.
- What you can do after
- Diagnose suspected deepfakes by fine-tuning ResNet50 to classify images as real or fake.
Read how Build a Text-to-Image Generator (from Scratch) explains deep fake →
Inside AI
A deepfake is digitally generated or altered media that realistically depicts a person or event in a way that is not authentic.
- How it is used
- Frames deepfakes as a harmful application of AI requiring vigilance because synthetic media can deceive individuals, organizations, and governments.
- How deep it goes
- Defines synthetic media and illustrates political, financial, and interpersonal harms without explaining generation or detection methods.
- Examples
- Uses fabricated political scenarios, corporate announcements, and a false Elon Musk video announcing Tesla bankruptcy.
- What you can do after
- Evaluate how convincing synthetic media can enable electoral manipulation, financial disruption, cyberbullying, and public deception.
Related concepts
- deep learninguses · Deep learning is a machine-learning approach that uses neural networks with multiple computational layers to learn representations from data.
- artificial intelligenceproduced by · Artificial intelligence is the field concerned with creating computational systems that perform tasks associated with human intelligence, such as learning, reasoning, perception, or decision-making.
- fabricated contentproduces · Fabricated content is media or information that has been artificially created or altered to appear authentic despite not representing genuine events or facts.
- generative modelingproduced by · Generative modeling is a machine-learning approach that learns patterns or a data distribution to generate new samples resembling the data.
- deepfake detectionaddressed by · Deepfake detection is the analysis of digital media to determine whether it has been synthetically generated or manipulated.
- generative adversarial networkproduced by · A generative adversarial network is a generative model with a generator and discriminator trained in competition, enabling the generator to produce samples resembling a target dataset or distribution.
These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.