GODDS: The Global Online Deepfake Detection System

NU 2024-251

INVENTORS

• Sarit Kraus*
• Venkatramanan Subrahmanian* (McCormick School of Engineering, Computer Science)
• Chiara Pulice
• Chongyang Gao
• Julian Baldwin
• Marco Postiglione
• Natalia Denisenko

SHORT DESCRIPTION

GODDS offers journalists a robust deepfake detection system that quickly analyzes audio, video, and image artifacts. It delivers comprehensive reports by combining multiple detection technologies with human review.

BACKGROUND

Fake audios, videos, and images are proliferating worldwide. Current detection solutions focus solely on the digital artifact and miss critical contextual insights. Journalists and media professionals struggle with unreliable verification methods that cannot adequately address high-profile cases.

ABSTRACT

GODDS is an online deepfake detection system that employs an ensemble of deepfake detectors alongside human review. It not only analyzes the submitted multimedia artifact but also derives contextual information related to high-profile individuals, organizations, and events. The system has processed artifacts from over 50 journalists at leading media outlets, validating its effectiveness in providing detailed and contextual deepfake reports.

MARKET OPPORTUNITY

The global market for deepfake detection was valued at $4.38 billion in 2023 and is projected to reach $42.63 billion by 2032, growing at a compound annual growth rate (CAGR) of 28.6%. This rapid growth is driven by the increasing accessibility of generative AI technologies and the widespread use of synthetic media in spreading misinformation, committing fraud, and influencing public opinion. (Source: MarketsandMarkets: "Deepfake Detection Market - Global Forecast to 2029").

DEVELOPMENT STAGE

TRL-5 Prototype Validated in Relevant Environment: GODDS has integrated its deepfake detectors with human review in a simulated journalistic workflow, successfully validating its key functions.

APPLICATIONS

• Deepfake detection: Verifies the authenticity of multimedia content for news outlets.
• Context analytics: Provides background information for enhanced digital content verification.
• Journalistic tool: Supports investigative reporting by reliably confirming content authenticity.
• Digital fact-checking: Assists media platforms in combating misinformation effectively.

• Provides comprehensive analysis: Combines automated detection with human oversight for robust verification.
• Enhances verification accuracy: Incorporates contextual information for more reliable results.
• Delivers rapid reports: Processes submissions quickly to support timely journalism.
• Addresses high-profile concerns: Focuses on cases with substantial background data to ensure thorough analysis.

• Sarit Kraus, Venkatramanan Subrahmanian et al, GODDS: The Global Online Deepfake Detection System, Proceedings of the AAAI Conference on Artificial Intelligence, 2025

IP STATUS
U.S. Patent Pending