Clarity, a cybersecurity AI startup, aims to protect organizations and individuals from the threats of deepfakes. By detecting deepfakes at scale and in realtime, the startup aims to enable users to avoid advanced social engineering and phishing attacks.
The project's main objective was to design Clarity Studio, a platform that allows users to upload media files—images, videos, and audio—to verify their authenticity and detect deepfakes. The platform also provides in-depth explainability for its analysis.
The project's aim was to design an innovative and transparent studio from scratch. The targeted opportunities were:
1. Incorporate high explainability
2. Prioritize interactivity
3. Gain user's trust in Clarity's verdict
4. Include customization
1. Performance: The studio should be able to handle high data volumes efficiently, minimizing delays or latency issues.
2. Accuracy: The solution should ensure accurate and reliable deepfake scores.
3. Customization: Users should have the ability to customize media files based on their specific needs.
3. User Experience: The studio should be intuitive, requiring minimal training, and should improve users' trust in the media.
As the sole product designer at Clarity, my role included:
1. Conducting competitor analysis
2. Iterating and exploring solutions
3. Writing documentation and organizing meetings to discuss explorations, pros, cons, and feasibility concerns.
4. Creating high fidelity mockups and prototypes.
5. Collaborating closely with developers to align design with backend capabilities
6. Overseeing the project handoff to the development team until completion.
Clarity Studio is a platform that allows users to upload media files—images, videos, and audio—to verify their authenticity and detect deepfakes. The platform also provides in-depth explainability for its analysis.
Deepfakes in digital media is rapidly evolving due to generative AI. In 2023, 15 billion images were created using AI. Deepfakes grew 10X between 2022 and 2023.
Clarity studio detects deepfakes. It can connect to any source of digital media and provides reliable AI detection.
The long-term vision for the studio is to become the universally preferred tool for deepfake detection. To achieve this, our initial user research focused on forensic experts. These experts sought a reliable tool to support their legal reports and provide credible evidence for court proceedings.
The key requirements, gathered from user interviews with forensic experts, are outlined below.
The studio must mirror aesthetics familiar to forensic experts.
Experts seek the ability to customize and add their expertise.
The studio must be fully transparent.
Experts want a highly-interactive studio.
A dark-mode studio to it align with forensic experts' technical backgrounds.
A Step-wizard to upload reports, add comments, edit score, and export media files.
Segment breakdowns providing key details, including Results Analysis, Detector Types, Detector Families, and Detector Levels.
Click and hover functionality for each segment breakdown, and for video and audio files, create an interactive timeline featuring a draggable marker.
I designed the UI library from scratch to support the studio's mission of being innovative, transparent, and trustworthy.
Below are some key UI components. Please note that the full UI library is not included here.
Following the design handoff to the development team, the studio is now live.
Since the studio's launch, thousands of media files have been uploaded by forensic experts, news organizations, and government agencies. The studio serves as a vital tool for preserving trust in the media.