ZeroFake
Design Challenge
Timeline
March 2025
48 Hours
Role
UX Researcher
UI/UX Designer
Art Direction
Tool
Figma
Deepfakes web
Midjourney
Runway
Chatgpt
ZeroFake is the first mobile app that leverages iOS Dynamic Island to deliver real-time deepfake detection, helping users instantly identify, understand, and respond to deepfake content, fostering awareness and trust in the digital age.
Timeline
15h 3/28-3/30
Research
Synthesis
4.5h
1h
40min
5h
2.5h
Ideation
Design
Deck Building
Challenge Prompt
Deepfake technology on platforms like youtube/social media poses a threat by spreading misinformation, defamation, and harmful content. The ease of creating and sharing deepfakes undermines trust, manipulates public opinion, and damages reputations. There is a pressing need for solutions to detect and prevent deepfake content on social media.
Solution
ZeroFake is the first ios app to provide real-time deepfake detection and use iOS Dynamic Island for instant alerts. With detailed reports, one-tap reporting, and interactive learning modules, it empowers people to spot, understand, and stop deepfakes, restoring trust in the content they consume.
Live Detection and Instant Feedback
On the landing page, users can begin the detection process by clicking the activate button, or by pasting a link or uploading footage.
Users can start browsing videos while lightweight AI models seamlessly analyze visual and audio artifacts. And they will be alerted instantly when the content is identified as deepfaked.
Users turn on the live detection activate button
*The live detection can also be activated on the control center interface once set up
Users can view detection status through dynamic island
When tapped, the Dynamic Island expands to reveal the detection status for the user
Instant warning appears when the current browsing content is identified as deepfaked to alert users
Users can tap to return to ZeroFake for a detailed analysis and take further actions
Detailed Analysis and Seamless Reporting
With detailed analysis featuring confidence levels and insights, users can understand the reasoning behind detection.
With one tap, the report will be forward to trusted platforms
After report submission, users can track the final decision, gaining a sense of achievement in contributing to online safety.
Interactive Modules for Deepfake Awareness
Quizzes to test and enhance users’ judgment
The learn feature provides users with educational activities, structured learning modules, and curated articles to build awareness and knowledge of deepfakes.
Process
What Is Deepfake?
Deepfake technology uses autoencoders and GANs to swap faces in videos. It extracts facial features from one person, maps them onto another using a trained model, and improves realism through a generator–discriminator loop. This process creates highly realistic fake videos that are hard to detect.
*Technical Mechanism
Autoencoders
A shared encoder compresses the source face into a latent representation capturing traits such as structure and movement, which a target-specific decoder uses to reconstruct the target face.
Generative Adversarial Networks (GANs)
The generator produces images from the latent space, while the discriminator judges their realism.”
Deepfakes on Social and Video Platforms
Deepfake Exposure vs Detection Accuracy
Distribution of Deepfake Content by Platform
Growth of Deepfake Videos (2019-2024)
While more than 60% of users have encountered deepfakes, detection accuracy remains low and many tend to overestimate their ability.
Deepfake content appears most on platforms like YouTube, X, and TikTok, with Facebook and other platforms also seeing an increase.
Deepfake content has grown rapidly on social media and video-sharing platforms, increasing 550% between 2019 and 2024 to 143,000 videos.
Social Risks and Impact
Impact on Media & Journalism
Misinformation & Trust Crisis
Reputation Damage & Personal Harm
Financial Fraud & Scams
Political Manipulation and Election Integrity
Legal and Ethical Concerns
Market Research
Who are the Target Users?
Young adults aged 18–29, who make up 37% of social media and video platform users, are the most active group online therefore particularly vulnerable to misinformation and the unintentional spread of deepfake content.
Key Opportunities
Get Suspicious Content Checked with Ease
Users like Emily often act impulsively, a seamless and fast way to verify content is essential to interrupt misinformation before it's shared.
Deepfaked Content Should be Warned or Reported
Without visible cues or reporting tools, users may unknowingly trust and amplify manipulated content.
Enhance User Awareness of Deepfake and Its Risks
Many young adults lack basic knowledge of deepfakes, leaving them vulnerable to manipulation and false narratives.
Ideation
During ideation, I brainstormed different solutions to address the pain points. After finalizing the key features, I explored the feature’s layout and user flow, establishing a foundation for the product’s structure.
Brand Direction
With a bold purple and neutral gray palette paired with modern, neutral typography, the brand creates a minimalist yet tech-driven aesthetic and conveys professionalism and trust.
Reflection
Technical depth
Although I researched deepfake-related technologies, mobile AI backend detection and Dynamic Island permissions. This was only a preliminary exploration, and further confirmation and expert consultation are needed.
User flows & features
Due to time limitations, not all flows and functions were fully developed. Future work should refine the various scenarios within real usage to ensure a seamless and logically consistent user experience.
Feedback & growth
Sharing the project more widely and collecting feedback from experts, peers, and potential users will help strengthen the concept and shape it into a more robust and impactful solution.