Taiwan’s first digital minister on building a trustworthy public system in the digital era

Low bandwidth is democracy’s biggest “bug,” according to Taiwan’s first Digital Minister and Cyber Ambassador, Audrey Tang. 

She was speaking at a fireside chat session at the GovTech4Impact World Congress 2026, held in Madrid, Spain, last May. 

As an advocate for participatory digital democracy, she noted that traditional electoral systems only capture public inputs in short bursts once every few years at the ballot box. 

This leaves the government ill-equipped to handle policy questions that are complex, contentious and rapidly evolving. 

This was the case in 2014 when a trade agreement between Taiwan and Beijing sparked a crisis, raising major concerns over cybersecurity, telecommunications and political sovereignty.  

Tang highlighted this moment, revealing the inadequacy of traditional democratic mechanisms in resolving technically complex and polarising issues. 

Taiwan turned that crisis into an opportunity.  

Its response became a prototype for a different model, one where large-scale public input was structured, clustered and translated into actionable consensus, rather than amplified into factional conflict. 

Tang contrasted a plurality-based governance model with the algorithmic outrage of social media, where systems are optimised to magnify differences into identity conflict. 

To prevent high polarisation, she proposed to make the most “bridging” voices more visible than the divisive ones. 

In redesigning the public sector, this meant that public servants shouldn’t just use digital systems to collect opinions, but architect these systems to identify overlapping areas, surface common languages that different groups can understand, and create conditions for co-creation. 

You may watch the recorded session here. 

Participatory models ahead of cyber threats 

Can participatory digital democracy models respond timely to cyber harms? 

Tang pointed to Taiwan’s deepfake financial scam advertisements, which used the synthetic likenesses of public figures to generate significant monetary losses. 

Since Taiwan sought to protect internet freedom and avoid broad content censorship, the government turned to structured democratic engagement to address this problem.  

“We sent messages to thousands of people asking what should be done, and a statistically representative mini-public of 447 participants then worked in small virtual tables to advance ideas that could persuade others,” said Tang. 

This exercise generated concrete policy proposals, such as joint liability for platforms hosting unsigned advertisements and technical penalties for non-compliant foreign social media platforms. 

Tang added that these proposals won strong public support, and were enacted into law rapidly – which was a move she linked to a reduction in deepfake scam ads. 

This showed that participatory digital democracy could be applied as a way of responding quickly to digital harms while preserving legitimacy. 

Addressing misinformation 

Digital democracy strategies often needed to guard against AI-powered misinformation that could fuel polarisation and unrest. 

During the Covid-19 pandemic, the country saw a rise of polarised claims around wearing the masks, centred around how masks were harmful or useless against virus spread. 

The government addressed this with the “Humour over Rumour” campaign, which communicated public health guidance using humour and virality. 

For example, a popular mascot was used to share a public message encouraging the public to wear masks and avoid touching their unwashed faces. 

Rather than relying solely on fact correction or shaming contending claims, the strategy also used civic psychology and platform design to make public health communication and accurate information more digestible. 

Tang noted that this type of messaging worked by reframing the issue around a broadly acceptable and behaviourally meaningful point, which saw an increase in handwashing and mask use.  

Exporting the model to other contexts 

Taiwan served as an exemplary model for digital democracy, which other governments have tried to adopt with different levels of success. 

Tang noted that poor adoption occurred when governments ran consultations without civic partnerships and without safeguards to reassure participants that they wouldn't be profiled or manipulated. 

“Governments shouldn’t attempt to import tools in a top-down way,” she warned. 

In other cases, digital democracy can scale when they are properly institutionalised, she said, highlighting the case of “Engaged California”.  

The US state developed a portal that invited local citizens to take part in conversations that impacted state governance, from wildfire mitigation to the social obligations of AI companies. 

This and Taiwan’s case showed that for these models to work, governments must support, institutionalise and partner with civic society, rather than displace their efforts.  

This in turn would enhance accountability, transparency, and public trust.  

AI as assistive technology 

For public servants who seek to achieve meaningful improvements using AI in a short timeframe, Tang recommended asking their local communities a simple question: “what sucks?” 

“Gather a broad inventory of everyday frustrations in public services, then use modern language models not only to summarise them but to close the loop with citizens in language they can recognise as their own,” she said. 

With large language models (LLMs), governments can now listen at the same scale they broadcast, said Tang. 

This allows governments to respond to people in ways that “reflect what they actually said, rather than reducing participation to a one-way extractive exercise.”  

Crowdsourcing solutions through hackathons could also help to scope and solve the problems that affect citizens the most, said Tang. 

If governments started with lived experiences to drive technological solutions, public legitimacy and practical service improvement could be pursued together, she noted. 

Ultimately, digital democracy was understood as a process that prioritises adaptability, answerability, and collective judgement, where human oversight and judgement remain critical. 

Warning that institutions should view technology as an assistive tool and not outsource their democratic muscles to machines, Tang highlighted that governments can be digitally-enabled, but human-centred.