On Day 1 of SusHi Tech Tokyo 2026, on the Main Stage, three of the most thoughtful voices in Japanese civic technology sat down to discuss how Japan might build the civic infrastructure to keep its democracy healthy through the next decade of generative AI.
The panel opened with a phrase Audrey Tang used in 2025 when describing Japan to the global civic-tech community: the last fortress. The framing reflects something real. Across the world, the share of people living in fully democratic countries fell from 51% in 2004 to 28% in 2024. Japan, for reasons worth examining carefully, has retained the civic conditions that make a recovery and a forward-going defense both genuinely possible.
This article is a read on what was discussed, who said it, and — most importantly — what makes Japan structurally positioned to build the civic infrastructure that other democracies will want to adopt next.
The Speakers
Mariko Nishimura is the founder and CEO of Heart Catch, an innovation consultancy that connects Japanese government, startups, and civil society. She moderated Audrey Tang's plurality session at SusHi Tech 2025 and is one of the most trusted bridges between Tokyo's policy circles and the global civic-tech community.
Ken Suzuki founded Digital Democracy 2030 — Japan's most ambitious civic-tech initiative to date — and previously founded SmartNews, one of Japan's leading news applications. He ran SmartNews's US operations starting in 2014 and watched the American information environment shift over more than a decade. Few people in the world combine a tech-founder track record with thirteen years of academic research on what he calls liquid society dynamics.
Haruyuki Seeki is the founder and CEO of Code for Japan, the country's leading civic-tech organization. He runs the Japanese implementation of Decidim — the participatory democracy platform that started in Barcelona — and partners closely with civic-tech organizations across Asia.
What Makes Japan a Strong Civic-Tech Foundation
Japan is one of the world's highest-trust societies. It has shared civic norms developed across centuries of continuous institution-building. The Edo period alone produced more than 250 years during which Japan deepened internal institutions, refined norms of conduct, and matured a coherent civic culture that has endured into the modern era. Japan is also a country where collective behavior in service of the public good has remained a default — visible in the way Japanese society coordinated through the COVID-19 pandemic, achieving strong public-health outcomes through shared norms rather than enforcement.
That underlying social fabric is what makes civic-tech work in Japan especially promising. Voting still functions normally. Civic participation has not collapsed into pure partisan identity. Citizens continue to treat the act of governance with seriousness. These are not minor details. They are the entire substrate that civic technology needs to be built on, and Japan has it.
Suzuki's Bridge — Adapting Plurality to Japan
The core of Suzuki's argument is that Audrey Tang's plurality concept — broad listening, structured deliberation, policy delivery — has been discussed in Japan for years and is now being operationally built at scale through Digital Democracy 2030.
Rather than trying to reform democracy in the abstract, Suzuki's project started with a specific, tractable problem: AI-generated scam advertisements impersonating Japanese politicians and celebrities have become a roughly ¥127.4 billion per year industry on major social platforms. The harm is concrete, the actors are economically motivated, and public awareness of the problem is already high. That makes it the right wedge for a civic-tech intervention. Build a working system for reporting and removing scam ads, demonstrate that the loop — citizens identify, platforms coordinate, policy follows — actually closes, and the same infrastructure extends naturally to harder problems over time.
The platform is in active coordination with the Japanese government for formal adoption — a meaningful step that civic-tech projects in most countries struggle to achieve.
Disinformation Is Not Misinformation
One of the sharpest distinctions of the panel — and one that anyone trying to think clearly about online discourse should adopt — came from Suzuki.
Misinformation is accidental. A reporter gets a story wrong. A scientist's preliminary data turns out to be flawed. A well-meaning person shares something they believed was true. Misinformation is undesirable, but it tends to be self-correcting in a healthy information ecosystem.
Disinformation is something else entirely. Disinformation is produced deliberately and is economically or politically motivated. Scam ads are economic disinformation — the operators earn revenue per impression. Coordinated influence operations are political disinformation — the actors are pursuing policy or electoral outcomes. The point is that disinformation actors are not confused or under-informed. They are responding to incentives.
This distinction matters because it tells you what interventions actually work. Media-literacy education works on misinformation — teach people to read sources critically and they will catch most accidental errors. Media literacy does very little on disinformation, because the actors are not trying to inform anyone and the audience is not the target of an attempt at truth. The audience is the target of an attempt at money or power.
The structural failure mode in any democracy is not that individual citizens fail to fact-check themselves enough. The structural failure mode is that we cannot reasonably ask individual citizens to constantly fact-check a barrage of professionally produced false content while running their actual lives. Not everyone has equal access to higher education. Not everyone has the time, the resources, or the institutional support to evaluate every claim that crosses their feed. Asking them to is a category error. It places the responsibility for democracy in exactly the wrong location.
The responsibility belongs further upstream — at the platform layer and the incentive layer.
Prebunking — The Vaccine Metaphor That Actually Works
The most exportable idea of the panel came from Haruyuki Seeki: prebunking. You vaccinate populations against future infection rather than treating each case after the fact.
The logic is simple. By the time a disinformation narrative has spread, the cost of debunking is asymmetric. The bad actor types out a false claim in twenty seconds. The fact-checkers spend two days assembling a rebuttal. By then the false claim has reached ten times the audience the rebuttal will ever reach. The economic and attention math of debunking does not work.
Prebunking inverts the timing. You teach people, before the attack, what an attack of this type looks like. You name the specific rhetorical patterns. You demonstrate the logical fallacies — the red herrings, the false equivalences, the manufactured urgency, the asymmetric burdens — so that when a citizen encounters the live version of that pattern, they recognize it. The recognition is the vaccine.
This is not hopeful theorizing. Community Notes on X works through a related mechanism in English-speaking markets, and Suzuki noted that Japanese-language Community Notes is currently under-developed because there are too few Japanese-language contributors. That gap is itself the opportunity Digital Democracy 2030 is positioned to address.
There is a new pressure that makes prebunking especially urgent right now. Translation friction between languages used to be a natural buffer against coordinated foreign influence operations. AI-powered translation has substantially reduced that buffer in the last eighteen months. Disinformation operations that previously needed native-language operators can now be run by anyone with API access. Many democracies are suddenly more exposed than they were two years ago — and the work being done in Tokyo right now is helping to build the defensive infrastructure that the rest of the world will eventually need.
The Incentive Problem — What Has to Change About Platforms
Right now, the dominant business model of major social platforms is impression revenue. Impressions are generated by engagement. Engagement is generated, disproportionately, by content that triggers strong emotional reactions. Disinformation is one of the most reliable producers of strong emotional reactions in the human nervous system. This is not a flaw of the platforms. This is the business model working exactly as designed.
You cannot solve the disinformation problem by asking individuals to be more discerning, and you cannot solve it by asking platforms to be more responsible if their responsibility comes at the cost of their revenue. The incentive has to be restructured.
Tech platforms in every country need to take responsibility — not as a moral position, but as a structural one. The era of platforms relying on a maximalist interpretation of free speech while profiting from the deliberate spread of false information is the underlying problem. Free speech is a value worth protecting. Profiting from monetized false rhetoric that hijacks public conversation is not the same thing as protecting free speech, and the two have been conflated for the convenience of the business model.
The alternative is to align platform monetization with information quality. That probably looks like a hybrid funding model — partially commercial, partially supported by governments or universities — where the economic incentive is no longer to maximize the engagement value of any content regardless of its truthfulness. It probably also looks like reputation systems where accounts that repeatedly amplify provably false information accumulate a visible record. People can still say what they want. The audience and the platform can still see who has been wrong before, and how often.
Japan, with its high-trust civic culture and its government's openness to coordinated civic-tech adoption, is one of the few jurisdictions where this kind of restructure has a real shot at being tried thoughtfully.
Critical Thinking as Civic Infrastructure
The skill that lets prebunking work, and the skill that platforms cannot ever fully substitute for, is critical thinking. Teaching populations to recognize patterns — not just facts — is the most durable form of democratic defense.
A debate where one side is constantly emitting false claims and the other side has to debunk each one in real time is not a debate. It is an asymmetric attack on the side that cares about the truth. Healthy civic discourse requires both sides to operate inside a shared framework of evidence, even when they disagree on values. When the framework breaks down, the system stops functioning as deliberation and starts functioning as a power struggle in which truth is irrelevant.
The civic infrastructure Japan is building is the kind that makes that asymmetric attack expensive again. Critical-thinking education that names the specific patterns. Access to facts that is structurally easy and structurally fast. A culture of civil debate that does not collapse the moment one participant decides to operate in bad faith. Tracking systems that let citizens see who has repeatedly spread falsehoods. All of this together is what gives a population the capacity to participate meaningfully in self-government in the age of generative AI.
The Decidim Case Studies — Why This Actually Works
Code for Japan's deployments of Decidim — the participatory democracy platform from Barcelona — show what structured deliberation produces when implemented seriously. Seeki described two cases briefly. In Barcelona, a community used Decidim to deliberate over funding for a cricket pitch — a minority sport in that context — and the structured process produced majority support for the minority initiative because the deliberation surfaced the underlying value of supporting community-building across cultural lines. In Kagawa, middle-school students were given a structured voice in city policy and successfully proposed and implemented a Mother's Day socks initiative.
These are small examples. That is the point. Civic-tech infrastructure does not need to start at the level of national elections. It starts at the level of a community deliberating over a cricket pitch or a city engaging students on a festival initiative. The patterns that work at that scale are the same patterns that scale up to national questions, once trust in the process has been built.
Why This Matters Beyond Japan
If Digital Democracy 2030, Code for Japan, and the broader Japanese civic-tech ecosystem succeed at building functional prebunking and deliberative infrastructure, the resulting playbook becomes a global export. Other countries will be able to adopt the same defensive infrastructure on a forward-going basis. Civic-tech communities in Asia and beyond are watching closely. Governments in Europe are starting to take the long view on what their information ecosystem needs to look like in 2030. Japan is well-placed to lead this conversation thoughtfully.
The "last fortress" framing was a respectful one. It is also an accurate one. The next two to three years will tell us whether Japan can build the infrastructure that turns the framing into a genuine global contribution to democratic resilience.
What This Means for Operators
For founders, builders, and operators reading this, the implications are practical.
First, the civic-tech category is genuinely live in Japan in a way it is not in most other markets. If you are building anything in the broader space — content provenance, prebunking infrastructure, deliberative platforms, reputation systems, fact-access tools — Japan is one of the few places where the government, civil society, and major foundations are actively coordinated and looking for adoption partners.
Second, the incentive-restructure conversation is going to happen somewhere over the next five years. The jurisdictions where it gets seriously prototyped will define the global template. Japan, with the substrate it has, is positioned to be one of them.
Third, the prebunking framework is exportable to non-civic domains. Marketing and growth teams should be paying attention to this work for entirely commercial reasons. The same patterns that vaccinate citizens against political disinformation also vaccinate customers against competitive misrepresentation. The pre-emptive objection-handling moves that the best B2B sales teams already use are essentially commercial prebunking. The civic-tech work being done in Tokyo is generating one of the most rigorous bodies of evidence on what makes this kind of intervention actually work.
The conversation at SusHi Tech was, in the best sense, hopeful. Japan still has the substrate. The infrastructure is being built. The exportable model might be three years away. That is not nothing. In the global democracy story of 2026, it might be everything.