Imagine someone spreading a fake video with the voice of a Moldovan politician asking for votes for a pro-Kremlin candidate. This is not an imaginary scenario. It is a description of what has already happened, in various forms, during the latest electoral years in Moldova. And yet, if you go on any major platform today, you will not find any label telling you that a piece of content was generated by artificial intelligence. The debate exists. The regulation does not.

For years, I have watched how Moldova treats deepfakes as a national security issue in political speeches. We are told the threat is real, that the state is paying attention, that the danger is taken seriously; and yet, the fundamental question remains unanswered: if this is a major digital threat, why has almost nothing been done to hold the platforms that profit from the spread of this content accountable?

Let's be clear: I am not arguing that labeling AI-generated content will solve disinformation. It won't. But it is a simple measure in the only sense that matters in public policy: it is clear, technically feasible, and shifts some of the burden onto the platforms that profit from spreading harmful content.

There is already a platform called UpScrolled, a small startup without billions in resources, which has publicly committed to labeling AI-generated content and removing synthetic media that risks misleading people. If a bootstrapped company can build the operational logic for labeling, then Meta, TikTok, and YouTube cannot seriously claim that this exceeds their technical capabilities. It is not an issue of possibility. It is an issue of political will.

The world is not standing still. The European Artificial Intelligence framework now includes transparency obligations for AI-generated or manipulated content, including deepfakes. China has gone further, mandating visible labels and embedded metadata for AI content, with penalties for non-compliant platforms. India has also tightened its rules, requiring major platforms to clearly label AI-generated content and retain traceability information. The argument that "this cannot be done" is no longer credible.

Someone will say that users can use VPNs to bypass any restriction. True in theory. But it is not a serious reason to do nothing. Digital literacy in Moldova still lags behind, and the people most vulnerable to manipulation are not, as a rule, the ones installing VPNs, reading policy updates, or checking metadata. If someone who doesn't know how to bypass a platform is even slightly less exposed to a flood of fake content, this is not a perfect solution, but it is still harm reduction.

The freedom of speech objection also deserves an answer. Let's not confuse freedom of speech with the freedom to spread fake content on an industrial scale. The first deserves protection. The second deserves regulation. A labeling rule does not ban political speech or satire. It demands provenance. It is a completely different thing.

Some will ask who Moldova is to tell Meta what to do. The answer is simple: any state that regulates its market, safety, or media has the right to set conditions for operating on its territory. Moldova does not have to pretend it is stronger than these companies. It just needs to be serious.

If our leaders can spend years talking about sovereignty, resilience, and resistance to Russian pressure, then they should be able to sit at a table with platform executives and say: these are the rules. Respect them if you want access to our users.

The real issue is whether Moldova wants to continue treating deepfakes as a talking point or start treating them as a governance problem. If the state believes that AI-generated manipulation is one of the main digital threats to the country, then it must stop outsourcing the burden of the response to the users least equipped to protect themselves.

Platform-level labeling is not the whole solution. But it is a practical step toward making disinformation more visible, less scalable, and harder to hide.

It is not censorship. It is basic digital hygiene. And at this point, Moldova should have moved past the stage of debating whether such measures are possible. The question is no longer whether platforms can do this. The question is whether the state has the courage to demand it.