Deepfakes
Also written: deep fake, Deepfake technology
Deepfakes are realistic fake videos or images, often showing people saying or doing things they never did.
Deepfakes are fake videos or images created using artificial intelligence. They can make it look like someone said or did something they never actually did.
For example, a deepfake could show a politician giving a speech they never gave, or a celebrity endorsing a product they never endorsed. These fakes can be very convincing, making it hard to tell what is real and what is not.
Deepfakes are synthetic media generated using deep learning techniques, primarily generative adversarial networks (GANs) or diffusion models. These models learn to map a source person's facial expressions and movements onto a target person's face in an existing video or image, or to synthesize entirely new video or audio content. The primary tradeoff involves the increasing realism and accessibility of generation tools against the potential for malicious use in disinformation, fraud, and harassment.
Deepfakes are a growing concern as AI becomes more capable of generating realistic synthetic media. Recent items highlight the challenges in distinguishing real from fake content and the development of technologies to combat misuse, such as watermarking and AI defense systems. The potential for deepfakes to mislead voters and impact elections is also a key focus.
- From deepfakes to DNA: the science of watermarking AIGoogle · Google DeepMind: The Podcast ·
- Helping voters navigate election information in the age of AIMicrosoft · Microsoft Source ·
- Reality is losing the deepfake warGoogle · Decoder with Nilay Patel ·
- We Got Slopped in 2025OpenAI · AI Inside ·
- Doppel’s AI defense system stops attacks before they spreadOpenAI · OpenAI News ·
- Elections and AI in 2024: Anthropic observations and learningsAnthropic · Anthropic News ·
- Anthropic Claude
- Anthropic has observed the role of AI, including its potential for misuse like deepfakes, in elections.
- Google Gemini
- Google is developing technologies like watermarking to track the origin of AI-generated content, including deepfakes, and prevent misuse.
- Microsoft Copilot
- Microsoft is working to help users navigate information in the age of AI, implying a need to address issues like deepfakes.
- OpenAI ChatGPT
- OpenAI has developed systems like Doppel that use AI to combat impersonation attacks, which can involve deepfake technology.
- AI Art
- AI art refers to images generated by AI, but not necessarily depicting r
- Voice Cloning
- Voice cloning creates artificial speech that mimics a specific pe
- Image Manipulation
- Traditional image manipulation involves altering existing ph
The term 'deepfake' emerged around 2017, combining 'deep learning' and 'fake'. It gained prominence through online communities sharing AI-generated pornography and political disinformation. Source ↗
You can explore AI image generation tools like DALL-E or Midjourney. While not creating deepfakes of specific individuals, experimenting with generating realistic portraits or scenes can demonstrate the underlying AI capabilities used in deepfake creation.
The Playbook: reusable how-tos, prompts, configs and tips pulled out of the news.
- PromptRole-play a student struggling with math for AI tutor simulationSimulate a student's math struggles and uncertainty for AI tutor evaluation. · 5 min
- TemplateReusable workflow templates for healthcare teamsTeams can create and share templates for common workflows like documentation or prior authorizations. · 15 min
- TemplateTemplate for building an AI tool tailored to your workA structured approach to defining and building a custom AI tool for specific business needs. · 15 min
- TemplateEvals CSV prompt list for agent skillsA CSV file template for listing prompts used in evaluating agent skills. · 20 min
- HeuristicDouble coding agent task completion with docs linksProviding coding agents with a documentation link significantly increases their task completion rate. · 2 min
Questions people ask
- How are deepfakes made?
- Deepfakes are typically made using deep learning AI models, like GANs, that learn to swap faces or synthesize realistic video and audio.
- Are deepfakes illegal?
- The legality of deepfakes varies by jurisdiction and how they are used. Creating them is often legal, but using them for fraud, defamation, or non-consensual pornography can be illegal.
First seen on Subvolts Feb 22, 2022. Updated Oct 8, 2026. Written by an AI model from the items above and reviewed against them weekly; report an error.