As generative AI-driven information manipulation threatens democracy, Taiwan stands at the forefront of global efforts to develop effective countermeasures.
On December 4, DSET’s Director of Democratic Governance, Kai-shen Huang, delivered a lecture at Brunei Gallery, University of London, titled “The Democratic Dilemma of Regulating AI-Generated Disinformation.” He outlined the challenges Taiwan faced during the 2024 presidential election due to AI disinformation and proposed potential solutions.
Huang highlighted several cases where generative AI was used to manipulate information. These included deepface audio falsely claiming Taiwan People’s Party Chairman Wen-je Ko had criticized the Democratic Progressive Party (DPP) candidate Ching-te Lai and fabricated videos linking Lai to the “Spring Breeze Project.”
Attackers have also used AI to generate fake avatars for fraudulent accounts. “Due to Taiwan’s unique geopolitical challenges, it has become one of the most targeted regions for foreign information manipulation, placing it at the forefront of this global battle,” Huang emphasized.
Balancing Disinformation Governance and Freedom of Speech
Another key issue discussed at the forum was how to regulate disinformation while safeguarding democratic values, particularly freedom of speech.
Huang explained that Taiwan’s strong commitment to free speech stems from its historical struggle for democracy following the martial law period. However, he also acknowledged a growing challenge: “Traditional regulatory frameworks struggle to keep up with the scale and speed of digital disinformation today.”
He further elaborated that past advocacy for free speech emerged in an era of information scarcity, where people sought knowledge and engaged in deep discussions. However, the current reality is one of information overload and shrinking attention spans, making traditional regulatory models ineffective against modern disinformation threats.
Identification Frameworks: A New Tool for Strengthening Governance
A key highlight of Huang’s speech was the introduction of frameworks like DISARM, which analyze cases of information manipulation by detailing the planning, execution, and dissemination strategies used. Using real-world examples, he demonstrated how attackers utilize AI-generated voices and distribute them via fake accounts.
Kai-shen Huang stressed that applying these frameworks helps standardize the process of identifying information manipulation and enhances discussions on governance, providing a clearer foundation for regulatory responses.
Audience Questions: Governance Strategies and Future Regulations
During the Q&A session, audience members raised insightful questions about whether Taiwan’s information manipulation efforts also encompass offline activities and whether AI can be used to counter disinformation.
Kai-shen Huang responded that information manipulation is not limited to online spaces—it often spreads through offline channels as well. Therefore, Taiwan must take a comprehensive approach that addresses both offline and online disinformation. He also noted that AI technologies are already being deployed to detect and counter disinformation, such as analyzing the authenticity of images or linguistic patterns. However, their use must be managed to avoid violating free speech or privacy rights.
Regarding the possibility of Taiwan reintroducing legislation similar to the Digital Intermediary Services Act, Kai-shen Huang explained that while previous attempts to pass such legislation were unsuccessful, they underscored Taiwan’s urgent need to govern disinformation effectively. He suggested that future proposals should prioritize transparency and multi-stakeholder discussions to ensure a balance between democratic values and effective governance, to secure broader public support.
Taiwan’s Global Significance in Combating Information Manipulation
As a resilient democracy facing external pressures, Taiwan serves as a critical case study in countering AI-driven disinformation. Kai-shen Huang concluded, “Taiwan’s experience highlights the necessity of international cooperation in tackling AI information manipulation. Our lessons can provide valuable insights for defending democratic values worldwide.”