Christina P. Walker,Daniel S. Schiff,Kaylyn Jackson Schiff
Abstract:This article presents the Political Deepfakes Incidents Database (PDID), a collection of politically-salient deepfakes, encompassing synthetically-created videos, images, and less-sophisticated `cheapfakes.' The project is driven by the rise of generative AI in politics, ongoing policy efforts to address harms, and the need to connect AI incidents and political communication research. The database contains political deepfake content, metadata, and researcher-coded descriptors drawn from political science, public policy, communication, and misinformation studies. It aims to help reveal the prevalence, trends, and impact of political deepfakes, such as those featuring major political figures or events. The PDID can benefit policymakers, researchers, journalists, fact-checkers, and the public by providing insights into deepfake usage, aiding in regulation, enabling in-depth analyses, supporting fact-checking and trust-building efforts, and raising awareness of political deepfakes. It is suitable for research and application on media effects, political discourse, AI ethics, technology governance, media literacy, and countermeasures.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is: How to systematically collect, organize, and analyze political - related deepfakes incidents to reveal their prevalence, trends, and impacts, and provide valuable insights for policymakers, researchers, journalists, fact - checkers, and the public. Specifically, the article introduces the "Political Deepfakes Incidents Database" (PDID), aiming to fill the gap in current research on causal evidence of political deepfakes and support research on the governance and countermeasures of political deepfakes.
### Specific Background and Motivation of the Problem
1. **Challenges Brought by Technological Progress**
- Deepfake technology has made remarkable progress in recent years, being able to generate highly realistic synthetic texts, audios, images, and videos, making it difficult for ordinary people to distinguish between true and false content.
- Use cases of political deepfakes include criticizing politicians, influencing voter turnout, spreading sexual - scandal accusations, and even influencing geopolitical events (such as Russia's invasion of Ukraine).
2. **Widespread Social Impacts**
- Deepfakes may lead to serious consequences, such as social unrest, political instability, the collapse of institutional trust, and the rise of irresponsible or authoritarian leaders.
- Research shows that deepfakes can influence elections through misinformation, exacerbate democratic regression, undermine the trust in media and government institutions, trigger widespread uncertainty and violent behavior, and damage public health.
3. **Limitations of Existing Research and Datasets**
- Although there are already some deepfake databases (such as the FakeCatcher database, AI - Powered Video Archive, Deepfake Detection Challenge Dataset), these databases mainly focus on technical detection methods and lack comprehensive coverage of deepfakes with political and social importance.
- Therefore, a comprehensive incident database is needed to provide detailed information on political deepfakes for systematic analysis and research.
### Goals of the Database
The main goals of PDID include:
- **Providing Public Resources**: Create a publicly available resource to assist research, practice, and governance work, especially in the areas of artificial intelligence ethics, technology governance, media literacy, and countermeasures.
- **Supporting Empirical Research**: By including the content, context, real - world connections of deepfakes, and relevant political and policy codings, provide a valuable resource to understand the prevalence, trends, and potential impacts of political deepfakes.
- **Promoting Interdisciplinary Cooperation**: Combine theoretical elements from political science, public policy, communication, and misinformation research to provide practical knowledge and support actions for a wider range of scientific researchers.
### Conclusion
PDID not only contributes to academic research but also can provide policymakers with insights into the scope and scale of political deepfake manipulation, helping them formulate targeted regulations and measures to protect the democratic process and national security. In addition, PDID can be used to develop and test new deepfake detection algorithms and study the impacts of deepfakes on public opinion, media trust, and other key political behaviors.