EU discusses banning deepfakes and AI-generated videos in official communications from public authorities

The European Union is discussing a ban on images and videos created by AI in official government communications. The initiative aims to combat deepfakes and attempt to restore trust in public communication at a time when political campaigns and international crises make any forgery particularly toxic.

Why the Ban is Being Discussed

In Brussels, officials fear that generated images and videos could be too easily passed off as real footage from government agencies. Against the backdrop of elections, harsh foreign policy rhetoric, and a constant stream of statements, any forgery can quickly spread through media and social networks, and then damage not only the reputation of a particular agency but also trust in official communications as a whole. The logic of officials is simple: if the state itself refuses to use AI-generated content, citizens will find it easier to distinguish verified information from manipulation.

However, the idea does not yet have final formulations. It remains unclear whether European authorities want to ban only deepfakes imitating officials, government buildings, and real locations, or whether they are talking about a broader restriction on any images and videos created by neural networks for political communication. This affects the scope of the future rule: it's one thing to block obvious forgeries, and quite another to abandon even neutral illustrations if they were created by a generative model.

What the Rule Could Look Like

So far, there is no ready-made law, but political and regulatory development is ongoing. EU bodies have already begun work on a practical code, which independent experts are to refine later. Then the possible ban will need to be agreed upon through the standard European procedure, and only after that can it become a mandatory requirement for official materials. If the initiative is formulated strictly, it could affect several levels of communication:

• Publications from EU institutions and press services
• Political messages and official statements
• Visuals imitating official figures or real locations
• Videos and images that could be mistaken for documentary footage

Even in current discussions, a broader intent is evident: European authorities want not only to close the specific loophole with deepfakes but also to reduce instances of harmful use of generative AI. Therefore, this story about official materials may be just one part of a more general approach to regulating synthetic content in politics and the public sphere.

Why There is Disagreement

The complete ban immediately faced critics. Their main argument is that the authorities themselves can remove AI-generated content from their channels, but this will not stop its spread on third-party platforms. The main stream of forgeries still flows through social media, messengers, and accounts that have nothing to do with official institutions.

If that's the case, a ban within government communications solves only part of the problem and barely affects the main source of disinformation. Many consider mandatory labeling as an alternative. Such an approach does not ban the technology entirely but forces clear indication of where synthetic material is presented to users.

For regulators, this is a more flexible option: you can reduce the risk of deception without abandoning useful AI applications. There is also a broader concern—if Europe responds to every new AI risk with the harshest restrictions, the region will begin implementing tools more slowly, which in the US and Asia are already becoming working standards.

What This Means

The EU is demonstrating that it is moving from general discussions about AI risks to specific rules for particular communication channels. Even if the final version turns out to be softer than a complete ban, the direction is clear: everything that looks like official information and can influence political trust will be regulated much more strictly than ordinary commercial or entertainment content.