Taiwan is renowned for its hardware prowess, but its software sector has historically been considered less prominent. Is it possible for this to change in the future? Additionally, what challenges does Taiwan face regarding its energy and rare earth mineral supplies?

That’s an excellent question. I began my tenure as Minister of Digital Affairs in 2022, the same year the generative AI revolution took off. During my time as minister, we witnessed a profound transformation in the software engineering profession.

Previously, becoming a software engineer required highly specialized training, particularly for maintaining large, pre-existing systems. This training was not homogenous; an hour spent on hardware-related software development, for instance, could not be repurposed for user-oriented client software. Because Taiwan’s hardware and semiconductor ecosystems offer exceptionally high salaries, they have historically attracted a vast amount of software talent to serve the hardware ecosystem, rather than consumers directly.

However, since the generative AI revolution of 2022, this has changed. Product managers, user experience designers, and those adept at translating needs into designs—even without extensive software engineering training—can now simply describe the software they want in English, French, or Mandarin. They can use what’s called “vibe coding”—that is, utilizing AI to generate the software infrastructure they need.

For example, we collaborate with recent university graduates whom we call “T-Ambassadors” (Transformation Ambassadors). Nearly 90% of these T-Ambassadors did not major in ICT-related fields, yet vibe coding enables them to rapidly translate requirements into viable applications, successfully pivoting their careers into the digital transformation space. Without a decade of software engineering training, they can build scalable applications with just a few dozen weeks of hands-on practice.

Furthermore, as I mentioned in my written response, we attract talent with a history of open-source contributions through our “Employment Gold Card” program. The digital domain is a burgeoning field, and to date, the Gold Card has attracted over 13,000 foreign professionals with global work experience to reside in Taiwan. Anyone who has contributed to the open-source ecosystem for eight years can obtain this four-in-one visa, which includes national health insurance and the right to start a business. I believe this is the future of the software industry: not spending a decade specializing in a single vertical, but being able to integrate into the open-source ecosystem and pivot quickly between different sectors, translating new market demands into vibe coding prompts.

Regarding the issue of electricity, I must emphasize that Taiwan has no intention of building a “super Stargate” to train massive models at a huge energy cost. Some are betting on hyperscale models, believing that if you train one with enough electricity, it will become self-aware—a single, giant entity that can perhaps solve all problems. That’s a bet some are making, but it’s not Taiwan’s path.

We believe, as I’ve said, that each industry and vertical should use tailored, moderately-sized, or “right-sized” models. In journalism, for example, you might need models for automated transcription and summarization. These models are small enough to run locally on your laptop, are not power-intensive, and don’t need to memorize every Hayao Miyazaki film. The point is that smaller, tailored models, especially open-source ones, are very energy-efficient and easy to fine-tune for your specific needs. Because of Taiwan’s dominant position in advanced chip manufacturing, we can enable this kind of Edge AI to run in an extremely energy-efficient manner. So, while we face energy challenges like any other country, I do not believe AI training will be our primary source of energy consumption, because we are not taking that brute-force, hyper-scaling path.

As for rare earth minerals and other raw materials, to my knowledge, Taiwan is actively working on finding substitutes and establishing efficient recycling mechanisms. Of course, we are more than willing to strengthen international cooperation on these fronts, but as this falls outside the scope of digital governance, I do not have detailed data to share.

Regarding the smaller, open-source models you mentioned, does this mean Taiwan will train its own proprietary models?

Yes, but we focus on training models for specific industries, not all-encompassing, general-purpose intelligence models. For example, we might train a cross-disciplinary model for quality control inspections in the manufacturing sector. But again, because it doesn’t contain content like movies, the model itself is quite small.

In language processing, we sometimes perform “distillation” and other times we fine-tune existing open-source models. I believe a very popular model in Taiwan is one based on France’s Mistral model, which was then fine-tuned with Traditional Chinese text to give it a better understanding of Taiwanese culture. So, the heavy lifting was already done by the Mistral team; Taiwan’s part was the localization, which is also very energy-efficient. And as you know, Mistral packs a great deal of capability into a very small model, so it not only saves energy but also democratizes the technology, as people can run it on their own phones and laptops.

One example is a project called “CyCraftGPT,” which is based on the Mistral model. It can understand cybersecurity incident reports and detect cyberattacks. It comprehends the Traditional Chinese used in these reports and can summarize the situation to coordinate defense efforts.

How severe are the daily cyber threats that Taiwan faces? Does the pressure from the PRC constitute a tangible danger?

We face millions of cyberattacks daily. This includes not just the security incidents we just discussed, but also information manipulation campaigns. Over the past year, there were approximately two million instances of information manipulation attacks. We know the PRC’s pressure on Taiwan is intensifying, with millions of cyberattacks and disinformation attempts daily, so the threat is very real.

I believe they are essentially testing our resilience. If we were to panic, if we failed to adapt and resist this manipulation and these cyberattacks—and cutting undersea cables counts as a denial-of-service attack—they would likely escalate their actions. So, my perspective is that this is a persistent threat. The only way we can defend Taiwan is by working with the international community to share threat intelligence, contribute to real-time joint defense mechanisms, conduct joint exercises, and so on, to make it clear to the aggressors that they cannot achieve any of their strategic goals, whether they be psychological, election interference, or otherwise.

Do you believe there is a high probability of a military invasion by the PRC?

I do believe that the outcome depends on ourselves. If we can ensure that the information manipulation, the attempts to sow division, and the narrative they want to project internationally—that “democracy is useless and only brings chaos”—do not gain traction, then I am confident the probability of military action will continue to decrease. So, this downward trend is, in fact, a reflection of our information and cyber resilience.

In Taiwan, do people feel a threat from Russia?

If you were to ask someone on the street in Taiwan if Russia has launched attacks against Taiwan, they probably wouldn’t say so. It’s not a prevalent topic in our society; we rarely talk about Russia. Of course, among cybersecurity professionals, we sometimes discuss Russian cyberattacks or related technical issues. We have also observed that the cyberattack playbooks used by Russia bear similarities to the methods we’ve seen from the Beijing regime. But in everyday conversation, the public is not very concerned with Russia, let alone any danger or threat it might pose to Taiwan.

What is the status of TikTok usage in Taiwan? What regulations has the government put in place?

When the Ministry of Digital Affairs was established in 2022, the first official document I signed extended the ban on products deemed a “harm to national cybersecurity,” such as TikTok, from the public sector to public venues operated or contracted by the government, like train stations. The use of such products is prohibited in these areas for cybersecurity reasons. Our definition covers any software, service, or hardware that is substantively controlled by a foreign adversary.

In the United States, as you know, there is an ongoing negotiation to transition TikTok’s US operations into a new entity owned by American shareholders, rather than being directly or indirectly controlled by Beijing. I am still involved with the Project Liberty Institute as a Senior Fellow, where I am following a proposal called “The People’s Bid.” My successor, Minister Huang Yen-nun, has stated that if the US version of TikTok successfully transitions to American ownership, he will direct TikTok’s operations in Taiwan to use the American version instead of the current international one.

Finally, I want to point out that the state of Utah just passed a law, effective next July, that allows any Utah citizen to migrate their social media accounts—from TikTok, for example—to other platforms like Threads or Bluesky, and TikTok must preserve all their original followers, content, and data. This is called the Digital Choice Act. Meta’s Threads already offers this kind of federated network, known as the “Fediverse.” If I post to the Fediverse from Threads, you can see it on Bluesky or other platforms without needing to log into Threads. This gives users a better choice in service quality, so they are not locked into a platform just because all their friends are there.

When I was minister, I specifically highlighted this as the way forward and created my Threads account, engaging primarily there. I am happy to report that over the past few years, the use of Threads among young people has grown substantially, far outpacing the growth of TikTok. In fact, in terms of traffic to Threads, Taiwan is the largest group. Even though we only have 24 million people, Taiwanese users outnumber even American users. This is a much healthier model, especially for young people, allowing them to use interoperable platforms without being forced onto TikTok.

So, for adults using their own phones and not within a government-operated space, there are currently no other restrictions besides those in the Anti-Fraud Special Act, which, for example, prohibits unsigned advertisements that could be part of a scam. In terms of anti-fraud measures, TikTok is subject to the same regulations as other platforms.

We heard the term “non-Red supply chain” for the first time today. How does this relate to hardware equipment from companies like Huawei? Also, some tech companies mentioned they chose to partner with France instead of Starlink because of Starlink’s close ties to the PRC. Could you elaborate on the “non-Red supply chain” concept and why Starlink was not chosen initially?

The concept of a “non-Red supply chain” is quite similar to the “Clean Network” initiative that began during the first Trump administration. The core idea is to ensure that all the different parts of a telecommunications service, from end to end and up to the application layer, do not rely on components directly or indirectly controlled by Beijing. This is not a new idea; the “Clean Network” was formally announced around 2020 and essentially stated a preference for trusted technology partners at every link in the supply chain. At the time, it was mainly focused on the telecom industry, but the concept has now extended to many other areas, such as drones.

As for my personal experience, when we were planning Taiwan’s communication resilience, we partnered with Europe’s SES and the UK’s OneWeb as our MEO and LEO satellite partners from the very beginning of the project.

Regarding why Starlink was not chosen initially, the reason is that when we started our planning in 2022, the Telecommunications Management Act at the time stipulated that if a telecom operator wanted to operate directly in Taiwan (rather than partnering with a Taiwanese telecom), its chairperson needed to have Taiwanese nationality. There were also restrictions on foreign shareholding ratios and the involvement of PRC capital on its board.

I recall mentioning this publicly in a recorded interview. When a business partner visited, I proposed an innovative idea: perhaps if we granted the chairperson—that is, Elon Musk—dual Taiwanese citizenship, it might solve the problem. I was actually quite serious, as we were also promoting the Employment Gold Card and an advanced program for naturalization for individuals with exceptional contributions to Taiwan, allowing them to retain their original passports. But this ultimately did not happen. At that time, Starlink also chose not to partner with a Taiwanese telecommunications service provider. That is why for the first phase of the plan, we chose OneWeb, as OneWeb agreed to partner with a local telecom, Chunghwa Telecom.

However, the situation is different now. In February of this year (2025), the Ministry of Digital Affairs opened up frequency bands to allow LEO satellite operators like Starlink, OneWeb, and others to use this spectrum. So if they were to apply now, I believe the ministry would review the application under the new regulations. But I don’t have specific information on when, where, or if Starlink is in the process of submitting such an application.

In the context of information warfare and disinformation, what do you consider the greatest threat to Taiwan’s democratic society?

If I had to single out the most dangerous message among the many, it would be this: “Democracy cannot deliver on its promises and only brings chaos.” In many democracies, because the social media environment favors engagement through polarization, you see more and more polarizing messages online. At the same time, many democracies also have populist politicians who harness these extreme voices to gain power, thereby feeding the authoritarian narrative that “democracy only brings chaos.”

This narrative can appear true in many democracies because they are becoming more polarized, populist narratives are gaining traction, and people do start to feel, “Oh, democracy really does just bring chaos.” It becomes a self-fulfilling prophecy.

According to the V-Dem Institute, Taiwan has been the top target of this kind of foreign information manipulation for the past 12 years. So by now, I believe we have become very resilient, like having antibodies; our immune system is very robust. But when I first joined the cabinet in 2016 as the minister in charge of open government, social innovation, and youth engagement, there was a deluge of information manipulation, and we didn’t have such strong “antibodies” back then.

For example, I became a cabinet minister in October 2016, but by September, rumors were already circulating that I could control GPS satellites and that anyone who turned on their phone’s GPS could be controlled and monitored by me. There were many variations of this rumor, like if you installed the anti-fraud software “Auntie Meiyu,” developed by our unicorn company Gogolook, that it was actually one of my tentacles for monitoring your conversations. This idea that “Audrey Tang can control your phone and other radio devices with her brainwaves” has been going on for nearly a decade. So, personally, I have been on the receiving end of this type of information manipulation for the past ten years.

If Taiwanese society can become immune to this information warfare, what is the “remedy”? What can the people of Taiwan rely on to counter it?

The remedy is “journalism.” By this, I don’t just mean professional journalists (who are, of course, very important) or professional fact-checkers in civil society (who are also crucial), but rather a “whole-of-society journalism.”

Specifically, starting with our curriculum changes in 2019, we shifted the emphasis from “literacy” (about consuming media) to “competency” (the ability to co-create media). According to the ICCS (International Civic and Citizenship Education Study) report, by 2020, Taiwan’s middle school students ranked among the top in the world for civic knowledge. They feel they can genuinely shape the public agenda and can collectively discover the truth through a journey of journalistic inquiry.

This has had a huge impact on “pre-bunking”—that is, anticipating what kinds of manipulation, like deepfakes or election interference, might appear in the future, and teaching the entire society to recognize these techniques. It’s the exact same principle as a vaccine: introducing your immune system to something that looks like the virus before you encounter the real thing. As a result, people can see through a deepfake video at a glance. About two and a half years ago, I personally deepfaked myself and showed everyone how it was done. The technology wasn’t very good then, but I told everyone it would get better and better. And indeed, in the election last January, deepfake audio and video did appear, but they had little to no impact on the outcome; in fact, they may have even backfired. I believe that prevention is always more effective than debunking after the fact.

This competency-based education empowers young people with a journalistic mindset. If you truly examine different sources, verify those sources, and construct a narrative, your mind becomes immune to manipulation and polarization.

You mentioned the experience of Switzerland earlier. Could you elaborate on this “participatory democracy” in the Swiss context?

I just visited Switzerland not long ago (from May 20th to 23rd) and spoke with many people working on democracy at the Canton level, as well as political consultants. I think the common thread they pointed out to me is that because people feel a high degree of political engagement at the state level (and at the national level, if they choose), there is less room for populism to enter the conversation. They said it’s not that Switzerland doesn’t have populist politicians, but they tend to become more moderate over time. Populism is simply less popular when the people feel they are genuinely part of the process and don’t need a strongman to protect them from some danger.

I believe this is identical to our experience in Taiwan. We also have populist politicians, but I think they too have become quite moderate over time. Furthermore, I believe “popular initiatives” are a hallmark of Switzerland, and we have our own citizen proposals on the national “Public Policy Network Participation Platform.” The ability for citizens to directly set the agenda is, in my view, an excellent antibody against populist politicians.

How does a “hacker” become a real-life government minister?

When I was 11, I lived in Saarland, Germany, for a year, near the French border. My father was pursuing his Ph.D. at Saarland University, and his dissertation was on the communication networks during the Tiananmen Square movement. He had been a journalist on the ground at Tiananmen until June 1st, 1989, which was fortunate for our family, as another journalist from his newspaper was injured in the square just three days after he returned to Taiwan. This memory is very personal to me.

I also remember how, in the living room of our home in Germany, my father’s research subjects—very young people in their twenties, exiled from Tiananmen—would come over to talk about democracy. Some of them eventually settled in Taiwan and taught at Taiwanese universities. And some of their students, in turn, were the ones who peacefully occupied Taiwan’s legislature in 2014.

I later became what is known as a “civic hacker,” which has nothing to do with cybersecurity. It’s about a movement called “g0v” (gov-zero), which creates open-source alternatives to existing government services. The basic idea of g0v is: for any government website (gov.tw) you don’t like, instead of protesting, you just build a better version (g0v.tw). By changing the ‘o’ to a ‘0’, you connect to a “shadow government” website that presents information, like the national budget, in interactive, visualized, and very engaging ways.

g0v started in 2012, and by 2014, I had already contributed to similar efforts for the multi-language version of the national dictionary and many other projects. But what really pushed g0v into the national consciousness was the three-week-long peaceful occupation of the legislature in March 2014. We ensured there was a livestream and a transcript of the dialogue, allowing the voices of about half a million people on the streets and many more online to be heard. We developed a tool where you could just type in your company’s name and immediately see the specific impact the trade agreement with Beijing would have on your business, enabling fact-based conversations. We helped over 20 civil society organizations facilitate discussions, and each day we were able to publish the consensus reached on the trade agreement from the previous day, as well as the points still under discussion. After three weeks, we had produced a very well-structured set of proposals, and the Legislative Speaker at the time, Wang Jin-pyng, simply said, “Alright, you win. We’ll adopt these. Go home.” We might have been the only occupation movement that became more cohesive over time, rather than more divided.

That is how, in 2014, I was invited to become a project consultant to the cabinet members of the time. After serving as a consultant for two years (because the rule for youth advisors—or reverse mentors—is that you have to be under 35), I became too old; I turned 35. That year, Dr. Tsai Ing-wen was elected president, and she promoted me from an intern consultant to the Digital Minister without Portfolio. In 2016, I started having my own youth advisors. My mission was to rebuild trust. In 2014, social trust was at an all-time low, polarization was severe, and President Ma Ying-jeou’s approval rating was just 9%. But by 2020, President Tsai enjoyed an approval rating of over 70%, partly because g0v had helped Taiwan successfully combat the pandemic that year without locking down a single city.

A French parliamentarian was very impressed with Taiwan’s model for combating disinformation, noting that beyond school education, the vitality of Taiwan’s civil society is key, as people do not rely solely on the state. Do you agree with this view? Also, can the information warfare from the PRC and Russia be compared?

In the past, a significant difference between European societies and Taiwan was the linguistic environment. The linguistic similarity across the Taiwan Strait is very high, so narratives from Beijing could be spread in Taiwan with only minor alterations and remixes, and people could easily understand them.

Precisely because it was hard to judge from content alone whether a message was a manipulation from Beijing or authentic, the work of discernment became increasingly difficult. In fact, due to the rise of AI and large language models, Europe is now facing the same challenge. In the past, you might have been able to judge authenticity based on content, but now that has become very difficult. In a sense, Europe’s situation is converging with Taiwan’s.

Therefore, we have shifted our defense focus from the “content level” to the “actor and behavior level.” For instance, last March, we held a civic deliberation on using AI to enhance information integrity. The main recommendation was that advertisements on social media—like Facebook, YouTube, and TikTok—must carry a digital signature, so you know who paid to sponsor the message. You cannot hide behind an anonymous bank account; otherwise, anyone with enough money could bypass civil society’s immune system and broadcast messages to a massive audience.

In response to this civic deliberation, Taiwan last year became the first country in the world to require “Know Your Customer” (KYC) for all online advertisements. If Facebook runs an ad without KYC and someone is scammed out of five million dollars as a result, Facebook is not only fined but is also jointly liable for that five million. After this law took effect in January of this year, if you scroll through Facebook or YouTube now, you will see almost no scam ads. There has been an overall reduction of over 90%, a remarkable success that did not come from banning content. Regardless of what an ad says, we insist that there must be a real person behind it, not half a million puppet accounts.

Because the KYC requirement clearly targets actors rather than content, and because we achieved a broad consensus through our alignment assemblies (where we sent 200,000 random text message invitations for public deliberation), we found that over 85% of the public—regardless of age, party affiliation, gender, ethnicity, or place of residence—supported these citizen-co-created measures. As a result, we did not face much pushback from the social media platforms.

So, do you think content moderation is still useful? For example, when someone makes an inflammatory statement, should the platform intervene?

What is unique about Taiwan is that we have the highest degree of internet freedom in all of Asia. We believe the way to counter authoritarian influence is not by taking away freedom of speech, but by adding more context, ensuring everyone learns journalism, and providing numerous tools like the Wikipedia-style collaborative fact-checking platform “Cofacts.” But you’ll notice that all these measures are simply about ensuring KYC—that the participants are real people, not bots. They add more context and allow you to clarify one-sided messages, but they never take any content down. In Taiwan, we basically do not practice content takedown.

Of course, this is not to say we don’t have any content-based laws; laws against content like child sexual abuse material still exist. What I mean is that in many of our laws, like the Anti-Fraud Special Act, we only encourage platform operators to restrict illegal “behaviors,” such as fraud. The focus is always on the behavioral level. If the behavior is illegal, it will of course be restricted. But as you can see, we’re talking about “actors” (e.g., impersonation) or “illegal behaviors,” not the “content level.” In Taiwan, the content level is almost secondary.

In 2017, I decided not to use the then-popular term “fake news.” I strongly disliked the term because both my parents are journalists, and I felt “fake news” was an insult to their profession. So, out of filial piety, I did not use the term or its translation. We use the Mandarin term for “disputed information.” So, we don’t debate whether content is true or false; we say, “Oh, this content is ‘polarizing.’” That’s very different. It’s not about the content itself, but about the “intent” behind it. Is it trying to sow discord? Is it trying to provoke polarization? Is it trying to incite hatred? But again, this is not a content-level conversation; it’s a behavior-level conversation.

(Observation from a journalist from Côte d’Ivoire) Before coming to Taiwan, I was surprised at how calm the people seemed in the face of the threat from the PRC. After arriving, I found this to be true. Having listened to your explanation, I’ve found the answer: Taiwan’s social resilience is incredibly strong. This “Taiwan Experience” should be exported to other countries.

Thank you for your kind words. In Taiwan, we have a saying: “Pressure makes diamonds.” Taiwan is a diamond of democracy, and we are very happy to export our experience.

As Taiwan’s first transgender minister, you represent the inclusiveness of your society. However, we are now seeing a regression on transgender rights in the United States. Do you worry that this “backlash” phenomenon could affect Taiwan?

I was born with naturally low testosterone, somewhere in between, but I’ve never let it bother me. My pronoun is “whatever.” I really don’t care. In Mandarin, the pronouns for he, she, and it—tā, tā, tā—are all pronounced identically. So, you don’t have to be forced to change your language to refer to me. I believe that absolute freedom of speech and mutual respect can absolutely coexist.

So, no, I am not afraid. I think there’s a situation in the US where each side demonizes the other, and only the most extreme voices from both sides go viral, so they have a very negative view of the other side. But I do believe there is a lot of common-sense, middle-ground thinking that can bridge different communities, as long as there is mutual respect. Wherever I go in the US, whether I talk to liberals or conservatives, they are all very respectful, and I really have not faced any discrimination. In Taiwan, LGBTQ rights are not a polarizing issue. And precisely because it’s not polarizing, the kind of populist backlash you mentioned is not a problem here, in my view.

My observation is that in Taiwan, there is a broad consensus within civil society and among the public about “Taiwanese identity” and the necessity of resisting attacks from the PRC. On the other hand, politicians and the media seem to be increasingly polarized. Is this polarization confined to the political and media spheres, without penetrating the rest of society? Would you agree with this analysis?

I completely agree. A decade ago, Taiwanese society was very polarized. You could see it in the use of flags. At rallies, you would see either the flag of the Republic of China or a pro-Taiwan independence flag, like the whale flag, but never both. They were in completely separate camps. But now, just this April at a rally, we saw both of these flags flying together in the same venue. That would have been unimaginable a decade ago. And as you said, they fly together because both sides are united against authoritarian encroachment on our freedom.

On a personal note, my father’s parents—my paternal grandmother was on the losing side of World War II, while my paternal grandfather was on the winning side. Obviously, it took some time for these two social groups to de-polarize. But today, Taiwan has one of the lowest levels of social, ethnic, and religious polarization in the world. So, even though, as you say, political polarization exists, it has not torn apart the rest of society. The fabric of trust is very strong between urban and rural areas, across generations, genders, social classes, ethnic groups, and religions. And because we are not polarized along these lines, mere political polarization cannot turn us into a populist polity, which leaves very little room for populist leaders to gain influence.

What does bubble tea mean to you?

To me, bubble tea symbolizes “open innovation.” Because it’s not patented, you can use white pearls, black pearls, black tea, green tea, rooibos tea—it doesn’t matter. You can mix it however you like, and no one will sue you. It’s all open innovation. I think this represents the dynamic nature of innovation in Taiwan: it’s open, it welcomes remixing, and it’s delicious. A world without Taiwan would be like a life without bubble tea—unimaginable.