The South Korean government is trying something unusual: before writing new rules for artificial intelligence, it is going to listen to ordinary citizens first. On the 12th of this month, the Ministry of Science and ICT and the National AI Strategy Committee will convene the first Public Deliberation Forum on AI Issues in Seoul, seating one hundred citizens chosen to represent a broad cross-section of the country. They will not be treated as spectators or passive recipients of expert lectures. They will be asked to deliberate, to weigh competing values, and to help shape the direction of national AI policy. The chosen topic for this opening session is one that touches almost everyone living through the age of generative AI: fake news created by machines. The government’s stated aim is to examine the many ways AI is affecting democracy and to discuss, from the perspective of citizens, the kind of society the country should build. The timing is not accidental. As the government weighs regulatory changes — including expanding the range of deepfake material subject to mandatory AI labeling, because synthetic content is now almost impossible to distinguish from the real thing — it recognizes that technical fixes alone will not be enough. There is a deeper public conversation to be had about where freedom of expression should end and where regulation should begin. Those one hundred participants may include teachers, nurses, farmers, factory workers, students, and retirees. Some will have strong opinions about AI; others may have never thought deeply about it. But they all share a recent experience: the unsettling feeling of watching a video and wondering whether it is real. The forum is an attempt to turn that collective anxiety into thoughtful policy. To support the effort, the government launched a public platform called Modu AI late last month, where citizens can submit ideas and respond to draft policies. The Korean word modu means “everyone,” and the name captures the spirit of the initiative. The future of AI is not something to be decided only by a few experts behind closed doors; it belongs to the people who will live with the consequences.
The first topic on the table is deceptively simple: where should society draw the line between creative freedom and harmful deception? Under South Korea’s current AI Framework Act, deepfake content must be accompanied by a notice or label making clear that it was generated with artificial intelligence. That sounds straightforward, but the law already contains an exception for artistic and creative works. In those cases, the label may be displayed in a way that does not interfere with the viewing experience. This creates a vast gray area that the citizen forum will now have to confront. Consider a realistic satirical video featuring a well-known public figure. It could be intended as parody, a form of expression that deserves protection. But if it is so convincing that people mistake it for an actual recording, it can quickly become a weapon of political disinformation. Similarly, synthetic content that carries no indication of its AI origin — a fabricated news article, a manipulated audio clip, an entirely invented photograph — sits somewhere between harmless fiction and deliberate fraud. The forum will consider whether such content should be brought within the scope of regulation, and where exactly the boundary should lie. Is it enough for a creator to say “this was meant as satire”? Or should the standard be based on how a reasonable viewer would understand the content? Should all AI-generated material be labeled, no matter how artistic or playful it is? Or should labels be required only when the content is realistic enough to deceive? These are not abstract philosophical questions. They affect everyone who scrolls through social media, shares a video with friends, or tries to make sense of a confusing political moment. The government is already considering expanding mandatory AI labeling to cover more deepfake material, and the citizen panel is meant to give that process a human anchor. The challenge is to protect the public without strangling creativity or turning every digital interaction into an exercise in suspicion. That balance cannot be achieved by technology alone; it requires judgment, and the government believes that judgment should come from the people themselves.
Beyond the question of labeling, the forum will take up the even harder question of responsibility. If an AI-generated fake goes viral and causes real damage, who should be held accountable? Under current law, the obligation to disclose that deepfake content was produced with AI falls only on AI businesses — the companies that develop or operate the underlying technology. That arrangement already seems too narrow to many observers. Content distribution platforms, social media companies, and individual creators all play a role in the spread of synthetic misinformation, yet they do not carry the same legal obligations. Even the existing requirement has been softened in practice. To reduce confusion in the industry, the government has suspended fines for violations until at least Jan. 22 next year. This means that, for now, there is little immediate penalty for failing to label AI-generated content. The citizen forum is expected to explore whether platforms should be required to detect and flag synthetic material, whether AI developers should do more to prevent their products from being misused, and whether individual creators should bear responsibility for the political or social consequences of their work. These are difficult questions, partly because accountability can be diffused across a long chain. The developer says it only made the tool. The platform says it only hosted the content. The creator says it was satire, or art, or someone else’s misunderstanding. And the viewer is left alone in the dark, trying to decide whether the thing they just saw actually happened. If the panel does its job well, it will not simply demand more labels. It will try to design a system in which responsibility is clear, transparent, and fair. This matters because regulation that targets only one link in a global chain is easy to evade. The panel’s conversations may reveal that a more comprehensive approach is needed, one that involves developers, distributors, and users in a shared system of accountability.
The urgency of these concerns has been forcefully expressed by President Lee Jae-myung himself. At a Cabinet meeting held at the Blue House, Lee warned that fake news, including AI-generated content, has become an especially effective tool for incitement. “Incitement driven by fake news is becoming a highly effective tool,” he said. He added that fake news goes beyond the level of an individual problem and causes such extensive social harm that a strong response is needed. The president’s words reflect a broader anxiety shared by many democracies around the world. Fabricated images and manipulated audio clips have already been used to influence elections, destabilize communities, and turn ordinary citizens into targets of anger and abuse. In South Korea, where digital platforms are deeply woven into daily life and online political debate is intense, the danger is especially serious. But the president’s call for a strong response creates an unavoidable tension. A democratic government cannot fight misinformation by simply suppressing speech. It must protect the openness of public debate while trying to prevent that debate from being flooded with counterfeit content. The answer, at least in part, may be public participation. When citizens themselves help define the boundaries of acceptable AI content, the resulting rules are more likely to be seen as community standards rather than top-down censorship. That is why the forum matters. It does not mean that ordinary citizens will decide everything alone; experts and officials will still have a role. But the conversations of these one hundred participants can inform government decisions and give policy a legitimacy it would otherwise lack. The president’s insistence on a robust response signals that the government intends to act. The question is whether those actions will be careful, principled, and supported by the public. The deliberation forum is a promising start.
This is not just a domestic exercise. South Korea’s experiment with citizen-led AI deliberation could become a model for other countries struggling with the same challenges. Generative AI has created a global crisis of trust, and no government has yet found a perfect answer. Some countries have proposed strict bans on certain uses of the technology; others have relied on voluntary self-regulation by technology companies. South Korea has chosen a third path: treating the question as a matter of democratic citizenship. The structure of the forum is important. One hundred citizens selected to represent a cross-section of the population suggests a deliberate effort to move beyond the usual voices — lobbyists, academics, tech executives, and political insiders — who dominate AI policy discussions. The participants will not simply be polled for their opinions; they will be asked to deliberate, to hear one another, to confront trade-offs, and to try to arrive at positions that serve the common good. There will be disagreements, perhaps sharp ones. Some will argue that any labeling requirement is an unnecessary burden on free expression. Others will insist that labels are not enough and that certain synthetic content should be removed outright. The value of the exercise is not that it will produce perfect unanimity. It is that society will watch a group of ordinary people wrestle with these disagreements honestly. That spectacle, in itself, could help restore public confidence in the fairness of AI regulation. The government has also launched the Modu AI platform to allow broader participation, meaning the conversation in Seoul is not sealed off from the rest of the country. Citizens across South Korea can submit ideas, raise concerns, and respond to policy proposals. The combination of a physical deliberative body and a digital participation platform may be the most innovative aspect of the entire initiative.
In the end, the South Korean government is attempting something both simple and profound: bringing the public into the heart of AI policy. The first forum on AI-generated fake news may not solve every problem, and one hundred citizens cannot possibly represent every perspective. Deliberation is not a substitute for strong legal safeguards or independent enforcement. But the willingness to ask is itself a step forward. In the age of AI, the greatest risk is not that machines will become too intelligent; it is that human beings will lose control of the shared reality that makes democratic life possible. If people cannot tell a real recording from a synthetic one, elections become vulnerable, public debate becomes chaotic, and trust in institutions begins to crumble. The citizen forum is an attempt to build a new kind of trust — trust in the rules that govern technology. It says to the public: your voice matters, your concerns are valid, and the future will not be decided entirely behind closed doors. The topic of deepfakes is only the beginning. Future sessions may take up even more complex questions, such as how AI should be used in healthcare, education, criminal justice, or national security. But the first step is the most important. By sitting together in a room and talking about what is true, what is false, what should be free, and what should be restricted, the participants will be doing the ancient work of citizenship in a brand new context. The technology may be unfamiliar, but the underlying need is timeless: to be able to trust what we see and hear, and to know that the people around us are dealing with the same reality. That is why the forum matters, and why the world will be watching. It is not just about fake news anymore. It is about whether democracy can adapt to the age of artificial intelligence.