Albania unprepared for the impact of artificial intelligence on elections, legal framework and monitoring mechanisms not yet adapted to the new digital reality

Artificial intelligence and manipulated content are penetrating Albanian elections, but the legal framework and monitoring mechanisms have not yet adapted to the new digital reality.

Artificial intelligence has already penetrated elections around the world. From fake audios of candidates to videos where politicians appear to say phrases they have never uttered, deepfake content generated with artificial intelligence is being used to attack political opponents, influence voter perception, or alter their electoral behavior.

Albania is no exception to this phenomenon. From the last electoral elections to the marathon anti-government protests known as the “Flamingo Revolution,” artificial intelligence is being widely used to create online content, often in its most harmful forms.

During the 2025 parliamentary elections, a number of disinformation campaigns were orchestrated through anonymous pages targeting political actors, without being restricted by digital platforms.

Other violations were also documented, including cases of unethical use of artificial intelligence, the creation and distribution of false content, or the manipulation of social media engagement through coordinated reactions or bots aimed at artificially increasing popularity.

But as the country heads toward new local elections in 2027, there is growing concern about legal gaps and technical capacities to address the misuse of artificial intelligence in democratic processes.

“Albania is not prepared for the risk that the misuse of artificial intelligence can cause,” says Megi Reçi, a digital rights researcher.

“The current Electoral Code addresses campaign coverage in traditional media, but bypasses the digital sphere. More specifically, the transparency of online political advertising, targeting practices, and oversight of online campaign financing remain unregulated,” Reçi added for BIRN.

The challenges are not only related to the lack of a comprehensive legal framework but also to limited institutional capacities.

The Central Election Commission, CEC, acknowledged to BIRN the legal gap as well as the lack of a dedicated protocol for identifying and handling content generated by artificial intelligence during electoral campaigns.

While assessing the need for additional guidelines and protocols, the CEC told BIRN that it was addressing the issue within a broader framework of existing competencies and cooperation with other institutions.

“The Electoral Code and the legal framework on media and campaigns provide general rules on electoral propaganda, electoral silence, and media responsibility, but they do not contain specific provisions that define, label, or sanction content created or manipulated with artificial intelligence,” the CEC said through spokesperson Drilona Hoxhaj.

“This is a gap that is also being observed in other countries in the region and the EU, and which requires addressing through targeted legal changes, not necessarily a complete revision of the Electoral Code,” she further added.

Forms of manipulation

Illustrative photo from Pixabay

The use of artificial intelligence is not problematic in itself, but it becomes so in cases where the public is not warned that a material has been created or manipulated with AI and when the content is used to create a false public perception.

Among the content produced with AI, deepfakes are intended for manipulation, which in the electoral context portray certain politicians as having said or done something that did not happen in reality.

During electoral campaigns, deepfake materials go viral much faster than they can be verified and addressed as false, which affects voter perception.

Erlis Çela, lecturer and disinformation researcher, says that the use of artificial intelligence in political content in Albania is still in a relatively early phase, but has a tendency to expand.

Çela also thinks that deepfake technology is not necessarily the most likely form of manipulation for the upcoming local elections.

Harder to detect, according to him, are contents that combine a real element with one or more manipulated elements, such as, for example, an old video presented as current, a statement taken out of context, a real retouched photo, or an audio where a few sentences are manipulated.

Synthetic materials in audio or video also pose another risk. In a polarized environment, a fake recording presented as a private conversation between politicians can be particularly effective if published a few days before voting.

“Manipulation does not need to convince the entire public,” Çela argues for BIRN, adding that behind it may lie several goals, such as, for example, reinforcing existing beliefs, discouraging opposing voters, or raising doubts about the integrity of the elections.

“The danger is not the false content in itself, but the effect it conveys through distribution mechanisms and problematic content, which is the manipulation and pollution of the information environment,” he emphasized.

Migen Qiraxhi from “Qëndresa Qytetare,” an organization engaged in monitoring electoral campaigns, says that the situation in Albania already constitutes a paradox.

While political communication has massively shifted to social networks, the regulatory framework and monitoring mechanisms have not at all kept pace with this rhythm, according to Qiraxhi.

“So, we have a paradox: while a large part of the campaign and its money passes into the online space, public control over this space remains almost in a manual version,” he emphasized.

“Artificial intelligence, manipulated content, artificial voice or image, and new forms of disinformation require a capillary response, from educational curricula to electoral legislation, to anticipate these phenomena before they turn into a massive problem,” Qiraxhi further stressed.

Legal gaps

State Election Commissioner, Ilirjan Celibashi, during a media communication. Photo: LSA.

The Albanian government aims to address some of the issues of artificial intelligence use through the draft law “On Digital Services in the Republic of Albania” – a country obligation to harmonize local legislation with the European Union’s Digital Services Act, DSA.

In July of this year, the draft law was published on the public consultation platform and has not yet been approved in the Albanian Parliament. The draft provides for obligations and cooperation mechanisms between state authorities and large digital platforms, and indirectly addresses some of the implications of artificial intelligence use, including negative effects on public debate and electoral processes.

However, this draft law does not specifically address the use of artificial intelligence and does not define a special regime for its use in electoral campaigns.

For Albania, which is a candidate to join the European Union, the issue seems more complicated than copy-paste adaptation with EU regulations.

“Large digital platforms, like Meta and Google, in countries like Albania operate mainly in a passive manner,” says researcher Megi Reçi.

“Due to the lack of legal obligations, the small size of the digital market, and the lack of dedicated staff with knowledge of the Albanian language and context, these platforms often fail to properly identify and label manipulated political content, political ads, or those sponsored by anonymous third parties,” she emphasized.

The Central Election Commission considers formal cooperation agreements with large digital platforms important, but also assesses that, starting from the next local elections, some key priorities need to be addressed.

These priorities, according to the CEC, relate to the legal definition of AI-generated or manipulated content and the obligation to label it as political advertising, as well as the creation of a rapid reporting and response system between the CEC, AMA, digital platforms, and law enforcement bodies, especially in the final period before voting.

The CEC also considers it important to strengthen its own technical capacities in identifying synthetic content, as well as voter education campaigns on the risks of AI-enabled disinformation.

“These are areas where work has begun, but which require time, resources, and political will to be finalized before 2027,” the CEC said.

For researcher Erlis Çela, the CEC should have a central role in cases where manipulated content is related to elections, but it should not remain the only institution.

“A single institution will never be sufficient in the face of such situations,” Çela said, while adding that, “AMA should have a role…, while platforms have information on how content is distributed and amplified.” /BIRN