// MODEL OPTIMIZATION AND PROMPT SYNTAX TERM
Deepfake Detection
The technology used to identify fake images, audio, or videos that have been realistically altered using AI.
TECHNICAL DEFINITION
Deepfake detection involves employing machine learning models, often convolutional neural networks (CNNs) or recurrent neural networks (RNNs), to identify subtle artifacts, inconsistencies, or statistical anomalies in media (images, audio, video) that indicate manipulation by deep learning-based generative adversarial networks (GANs) or autoencoders.
BACKGROUND
Generative artificial intelligence (GenAI) is a subfield of artificial intelligence (AI) that uses generative models to generate text, images, videos, audio, software code or other forms of data. These models learn the underlying patterns and structures of their training data, and use them to generate new data in response to input, which often takes the form of natural language prompts.
READ MORE ON WIKIPEDIASYNONYMS & ALIASES
- Media forensics
- Synthetic media detection
- AI-generated content detection
USAGE NOTE
Deepfake detection is crucial for verifying the authenticity of digital media in an era of advanced generative AI.
DEVELOPERS
Organizations developing technology related to Deepfake Detection.
Sensity AI specializes in visual threat intelligence, including technology for detecting sophisticated deepfakes and manipulated media across various digital platforms.
Microsoft has been at the forefront of AI research and development, including creating tools and participating in initiatives like the Content Authenticity Initiative (CAI) to detect deepfakes and ensure media authenticity.
Meta's AI research division actively conducts and publishes research on deepfake detection, hosts challenges to advance the field, and develops technologies to combat misinformation and manipulated media on its platforms.
Google's Jigsaw unit develops technology to address threats to open societies, including tools and research for detecting deepfakes and other forms of synthetic media designed to mislead or misinform.
Reality Defender offers a platform for comprehensive deepfake detection, providing APIs and enterprise solutions to scan and identify AI-generated or manipulated content across various media types.
Truepic provides verifiable media solutions, including technology to authenticate visual content at the point of capture and detect deepfakes or manipulation, ensuring the integrity and trustworthiness of images and videos.
Pindrop specializes in voice security and authentication, developing AI-powered technology to detect sophisticated voice deepfakes and synthetic audio used in fraudulent activities and identity theft.
As a founding member of the Content Authenticity Initiative (CAI), Adobe is actively developing tools and standards to embed content credentials and detect manipulation, including deepfakes, within creative workflows.