Montenegro is not ready to deal with artificial intelligence used in the media.
At the end of last year, the then director of the Podgorica Health Center, Danilo Jokić, was the target of fraudsters on Facebook, who, with the help of deepfake, i.e. artificial intelligence (AI technology), misused his personality on the now unavailable Heart Wellness page to sell dubious medical preparations for cleaning blood vessels and treating rheumatism. Mr. Jokić reported this fraud to the police, i.e. the cybercrime inspectors, and then it was announced that those that suffered the most harm were the citizens who may have fallen for the fraud. This is just one case of abuse of artificial intelligence in Montenegro that indicates the consequences that can be caused by the unregulated issue of using new technologies. AI technologies also open up various ethical dilemmas, primarily in terms of how they are used in shaping media content.
However, this issue, although it trending highly globally, has not yet been recognized as important in Montenegro. A set of media laws has been pending adoption for a long time, and the drafts prepared so far do not properly treat the many ongoing challenges in the media environment, such as, for example, the AI technologies. At the end of 2023, Montenegro adopted Media Strategy for the period from 2023 to 2027, but this document again fails to specify how to deal with this phenomenon.
,,While there is a lot of talk about artificial intelligence at the European level and its influence on the work of the media outlets, not only has Montenegro not even identified this possible issue, but we have not yet resolved the issue of social media and the regulation of portals. Although it is commendable that we have adopted a Media Strategy last year, it is sufficient to say that in that document, which maps all the challenges for the media community, there is not even a word about artificial intelligence", said the Media Trade Union of Montenegro (SMCG) for Res Publica.
At the end of last year, the Council of Europe adopted Guidelines on responsible use of artificial intelligence systems in journalism, which address ethical dilemmas and the requirement to inform users that such systems are used in the work of the media outlets. The use of AI technologies is not disputed in that document, because, as it is stated, "such technologies, in accordance with professional ethics and human rights, can contribute to the resilience of journalism in the digital age".
The AI Act was also recently adopted at the level of the European Union (EU) which, although it is not strictly related to media outlets, contributes to the regulation of this area. The EU takes more and more seriously the fight against fake news, so the European community recently sent a request to technology giants such as Google and Meta to clearly identify all content that is generated by artificial intelligence - be it texts, photos or any other media materials.
Some countries, such as the USA, Japan, Great Britain and China, have prepared guidelines, regulations and laws regulating the use of AI technologies, and it seems that Montenegro is still far from making such move.
Only one Montenegrin newsroom uses artificial intelligence
So far, only one newsroom in Montenegro has introduced an artificial intelligence system that helps it in its daily work. The DAN portal has, since the middle of 2023, virtual presenters who read the information prepared by the editorial staff of the portal. The presenters are called Ana, Sara and Mara AI, and it is the first such project in Montenegro. It works as follows: first the editor making a selection of news and photos, and then the artificial intelligence processes the news. Using the TTS system, text is turned into audio, and then this audio is inserted into the system that starts the virtual presenter, which is also the basis for her movements. At the end, everything is checked by an editor, a journalist and a trained operator.
The Media Trade Union of Montenegro advocates that the use of these systems should always be done in responsible manner. "It should be motivated by a real desire for innovation, and not by mere entrepreneurial goals to reduce the workforce.", says the Media Trade Union.
Artificial intelligence has a dual nature. On the one hand, the authors of "Artificial intelligence and communication: A Human–Machine Communication research agenda” Andrea Guzman and Seth Lewis, say that AI technologies can improve the media economy, personalize the content but can also improve the production processes. For example, they can create transcripts and infographics or filter comments that contain illegal speech on portals, which is something that the media in Montenegro are largely unable to deal with.
On the other hand, the text “The Impact of Artificial Intelligence on Media, Journalists, and Audiences" published on the open scientific platform Frontiers, warns that artificial intelligence has a multi-layered impact on the media outlets, journalists, audiences and media content and that it can additionally affect the vulnerability of the media outlets and their dependence on large technological platforms. For certain things, such as presenting a wider context or checking facts, a human involvement is necessary.
In the world of media, ChatGPT and Sora are becoming increasingly popular. These are AI technologies that have been developed since November 2022 by OpenAI, an American laboratory where artificial intelligence is researched. ChatGPT is a chatbot that can, among other things, write journalistic texts on a given topic. Sora can produce complex videos that are up to one minute long. "The model understands not only what the user requested, but also how these things exist in the physical world," the official website states.
And as useful as these technologies can be, they should be used with caution, as AI algorithms can easily be abused to spread fake news, disinformation, and other illicit content.
Source: blink.ba
This is best illustrated by the case from 2022, when Microsoft's MSN news site, which previously fired dozens of journalists and editors and replaced them with an AI algorithm, published a bunch of fake news. It turns out that MSN's algorithm took a particular liking to Exemplore, a comical website that mainly features news about paranormal and various conspiracy theories.
Deepfake dilemmas
One of the currently most famous AI technologies that is often used to spread fake content is deepfake. These are fake videos created by artificial intelligence by replacing or digitally altering someone's face and voice.
"Deepfake can sometimes be useful, for example when it is used to “bring into life” some historical figures or extinct living creatures, for the purpose of scientific interest or when used for private or entertainment purposes, so thanks to deepfake you can see how you would look like in your favorite movie. However, practice shows some shortcomings of this technology, as well as ethical problems," says journalist Jovana Damjanović for Res Publica.
According to her, deepfakes are used to manipulate the public and, as such, they can permanently ruin someone's career, private life, and integrity. "The question is how well media workers are trained to recognize a deepfake, and therefore to stop its spread and the instrumentalization of the media outlets, but also how general is the knowledge about a deepfake and how difficult it is to prove that it is not a real person", says Damjanović.
The recent case of spreading explicit photos of singer Taylor Swift on the social network X, which were actually deepfakes, also showed how far the global legislation generally lags behind the development of the AI technologies. Her photos, although they were taken down from X very quickly, were seen and spread by millions of people.
There are also sites where you can find deepfake porn videos of celebrities from the world of art, sports, politics... One such site is MrDeep Fakes, registered in 2018, which has millions of visits per month. The victims are mostly women. The content on this website is free, and you can also order videos with a specific celebrity, which is paid in cryptocurrencies. It even has instructions for the creators on which deepfake programs to use.
There are also such cases in the Balkans. In February of this year, the public in Serbia was informed that several 13-year-old boys were suspected of using AI applications to create explicit images of their female friends and teachers in a Belgrade school and then sharing them further.
Some countries have laws that help them fight deepfake abuse. The Deepfake Act adopted in California criminalized deepfake pornography without consent and gave victims the ability to sue. This Act also prohibits the use of deepfakes during election campaigns.
The Online Security Act of the United Kingdom in 2023 made it illegal to share explicit images or videos that have been digitally manipulated. However, this Act only applies in circumstances where the victim's distress has been intentionally or recklessly caused, and the victim can seek compensation relying on some criminal laws, all of which can be very complicated and difficult to prove.
In the EU, the use of deepfake is regulated by the Artificial Intelligence Act. Although it does not impose a direct ban, the Law regulates such content in such a way that the creator is required to indicate that it is a deepfake content.
About 40 percent of Montenegrin citizens do not understand the term "artificial intelligence"
According to a research carried out by the Damar Institute in 2023 titled "How the citizens of Montenegro perceive artificial intelligence", six out of ten citizens of Montenegro, or about 60 percent, have heard of the term "artificial intelligence".
However, only six percent of the respondents understand this technology very well, 25.6 percent understand it partially, 28.3 understand it poorly, and even 40.2 percent do not understand at all what exactly is meant by this.
The Damar Institute explains that there are a number of factors why the results are so diverse, and among them are the lack of education, the difficulty of keeping up with the latest advances in the field of artificial intelligence, fear of technology and the unknown...
The research titled ”Weak Mechanisms and Outdated Code of Ethics - Self-Regulation of Media Outlets in Montenegro" carried out by the Media Institute of Montenegro (IMCG) states that there was no intensive discussion on the impact the algorithms have on the work of the media outlets and of their visibility on the social media, and there was also no discussion on the role of artificial intelligence in online media outlets. The research ascertains that "Only occasional texts were observed in which the Montenegrin media reported on cases of false information that were created with the help of artificial intelligence, texts in which they tested ChatGPT, and asked journalists what they thought about the impact of the artificial intelligence on their work."
The EUROPOL Report from 2023 warns that, within a few years, up to 90 percent of the online content could be synthetically generated.
"We hope that in Montenegro we will finally start to act in accordance with the moment in which we live," is the conclusion of IMCG.
What does the EU AI act foresee?
The AI Act of the European Union (EU) is the first comprehensive legal framework on artificial intelligence in the world. It guarantees the safety and basic rights of people and companies, while at the same time supporting the development and application of artificial intelligence.
The Law rests on various rules for AI systems that are based on a certain degree of risk, i.e. one out of three.
The vast majority of AI systems, such as the systems for recommendations or spam filters, fall into the category of minimal risk and shall not be tightly controlled.
Artificial intelligence systems that are identified as high-risk are regulated according to strict requirements, and examples of such systems are those used in health, law enforcement, border control...
Artificial intelligence systems with unacceptable risk are prohibited. These are systems that manipulate human behavior, such as, for example, some biometric systems for recognizing emotions. The EU AI Act also defines a specific risk of transparency. Deepfakes, chatbots and other similar content generated by artificial intelligence will have to be labeled as such.
Montenegro, which aspires to become a member of the EU, must understand the seriousness of the artificial intelligence phenomenon as soon as possible, especially in the field of media outlets, and then work toward reaching EU standards in this area.
The blog was created as part of the “Tales from the Region” initiative led by Res Publica and Institute of Communication Studies, in cooperation with partners from Montenegro (PCNEN), Kosovo (Sbunker), Serbia (Autonomija), Bosnia and Herzegovina (Analiziraj.ba), and Albania (Exit), within the project "Use of facts-based journalism to raise awareness of and counteract disinformation in the North Macedonia media space (Use Facts)" with the support of the British Embassy in Skopje.
Please refer to the Terms before commenting and republishing the content. Note: The views and opinions expressed in this article are those of the author and do not necessarily reflect the views of the Institute of Communication Studies or the donor.
