How newsrooms in BiH and the region are using AI tools.

Illustration: Said Selmanović

Artificial intelligence tools are widely used in newsrooms across the world — for transcribing interviews, translating, generating short news items, and analysing data.

In Bosnia and Herzegovina, however, only a few newsrooms have integrated AI tools into their daily workflow or invested in tools tailored to their specific needs.

Dinko Dundić, editor of the news portal Fokus.ba, says that their journalists have been using ChatGPT for a while now, and have recently begun experimenting with other AI tools such as Midjourney and Ideogram for document analysis, SEO optimization, and infographic creation.

He explained that AI tools save time, especially when it comes to translation and data analysis, and that they are becoming increasingly advanced, useful, and reliable in journalism.

“Of course, that doesn’t mean we can take everything they generate at face value. Journalists still have to verify everything in the end. Traditional fact-checking is mandatory for all content created with the help of AI tools”, he emphasized.

Journalists at the Bosnian newsroom of Bloomberg are experimenting with various AI tools, primarily using ChatGPT.

They are satisfied with the tools’ efficiency, but their priority is ensuring that the journalist using AI tools has strong subject-matter knowledge, says Rijad Durkić, editor of Bloomberg Adria Digital BiH.

“Above all, I’m referring to the use of AI in advanced searches. The time savings are certain, and I believe that on a global scale, AI also reduces the need for human labour on a macro level, which raises many questions. All of these tools have facilitated and improved journalists’ work, especially in specialized media. The benefits are numerous, particularly in analysing lengthy reports and large documents,” Durkić said.

The newsroom of the portal Klix has integrated AI-based software into its daily operations. Their solution for using AI to combat hate speech in comments even received a regional award.

“In addition, we use the TurboScribe tool for transcribing audio material. We have also integrated certain tools that help us create images or share news on social media”, explains Klix.ba editor-in-chief Semir Hambo.

The effectiveness depends on the tool, and not all tools perform at the same level, according to the experience of the Klix newsroom.

They are aware that, despite the many benefits of using AI tools, the human touch remains irreplaceable.

“These tools, in any case, are not yet at a level where they can complete tasks without human intervention”, Hambo concludes.

Artificial Intelligence Cannot Function Without Human Text

Many journalists and editors avoid using artificial intelligence tools, believing them to be neither reliable nor useful.

Sanja Vasković, editor of the investigative portal Spin.info from East Sarajevo, says that their newsroom is not in the habit of using AI.

“We deal with investigative journalism, and I’m not sure how much it can help us in that field. Interestingly, we recently asked ChatGPT about the construction of a certain project, and it quoted parts of a previously published article of ours. I believe that for colleagues involved in daily news reporting, it can be a useful tool”, says Vasković.

Science journalist and editor of the Nauka govori portal, Jelena Kalinić, uses AI tools moderately because they do not suit her needs.

“We use Canva (which is not fully AI but has AI features), and tools for image creation. The images we use as cover visuals for articles are never 100% AI-generated — they always have the touch of the person who designs them, including some of our own solutions and templates, and selected colour palettes”, says Kalinić.

She doesn’t use large language models like ChatGPT but believes they are good allies in daily newsroom operations.

“For those of us who do deep analysis, interviews and investigations, opinion pieces, and explanatory texts — these tools are nearly useless. They’re good when we need a quick definition so we don’t have to retype text, but our articles need richness, analysis — and chatbots produce generic content with a forced tone, often repeating annoying phrases like 'fascinating,' or 'did you know?'. Personally, I find these models very unsuccessful even when it comes to writing headlines and subheadings — even when you prompt them, they fail to recognize a strong moment in the text, a good quote, or humour”, Kalinić explains.

She also avoids using such tools for editing texts and headlines, as that would mean the chatbot needs to be fed with the original article.

“Of course, someone else could do that with our article, but at least we don’t want to be the ones to give that human content to AI. Because AI cannot function without human text — otherwise, as research has shown, information becomes degraded if AI is fed with AI-generated content.”

For her, the most useful AI tools are transcription programs such as Otter.ai, especially for English, though she notes that even these require verification.

The Need for Education

Helena Javor Ibrahimbegović, a news editor at Face TV, says their journalists do not use any artificial intelligence tools.

“Based on the experiences of our colleagues from other media outlets with whom we’ve discussed the use of AI tools, and from some of our own trial experiences, we’ve concluded that these tools are efficient but not reliable. Free versions, which are most commonly used, are especially unreliable because they are not regularly updated. They often provide inaccurate information — for example, the same person may be assigned different roles across several generated texts”, explains Javor Ibrahimbegović.

Journalists from the newsroom of the portal Impuls have not used artificial intelligence technologies. They do not consider them inherently bad, but until their use is regulated, they won’t trust the accuracy of such tools.

“Precisely because of their widespread availability, I think it’s easy to generate and distribute false information. For investigative articles that require time and cross-checking from multiple sources, AI doesn’t seem like a method we could use”, says Impuls editor-in-chief Jelena Jevđenić.

She is not familiar with all the tools and believes that adopting new technologies takes time.

The editor of Fokus emphasizes the need for continuous learning and adaptation so that media and journalists can remain competitive in the profession.

“Just as journalism changed with the arrival of the internet and later social media, there will undoubtedly be major changes as AI becomes more widely used in journalism, but also in life in general”, he says.

Feđa Kulenović, assistant professor at the Faculty of Philosophy in Sarajevo, believes that the use of AI in newsrooms in BiH is currently in an experimental phase, and he notes that this pattern was similar when social media first emerged.

“Not just here—the strategic foundations for serious integration were established quite late. In this new digital transformation, several things are becoming clear: media outlets lack the human capacity and knowledge to know what to do with this new technology, and they risk falling into the trap of working for others without any compensation”, says Kulenović.

He observes that strategies necessary for significant progress are only just beginning to develop, and he is confident that in the next few years we will see serious, ethically grounded efforts to better harness the potential of AI.

“Of course, all of this depends on financial resources, which will most likely slow down rapid growth.”

As a media expert, he recognizes that artificial intelligence can—and already does—play a useful role in journalism. There is a wide body of scientific studies and groups like JournalismAI that highlight its benefits from various perspectives, he adds.

“First and foremost, AI helps shape content more effectively for specific target audiences, which leads to even greater personalization—but always keeping in mind that a human is at both the beginning and the end of the process”, Kulenović concludes.

Newsrooms in the Region

In newsrooms across neighbouring countries, AI tools have been used for years. The portal Dan introduced the first AI news anchor several years ago, and newsrooms across the Balkans use artificial intelligence technologies in their daily operations.

Jure Tepina, Head of Digital Content at CME Adria (net.hr, 24ur.com), says that these two portals from Croatia and Slovenia have been using AI for a while now to automate journalistic tasks. Throughout the process, human oversight of AI operations is mandatory.

“That’s why we use tailor-made solutions or have developed tools with our own programmers. This way, we maintain maximum control over which sources are used — controlling data entry points is essential to prevent built-in bias”, says Tepina.

AI-generated content is treated like any unverified source.

“The results so far are still mixed—we’ve had many hits, but also many misses”, Tepina added.

The newsroom of the portal Blic.rs has been using ChatGPT for a while, along with several internally developed AI tools that assist journalists with the technical aspects of preparing texts—but not with writing itself.

Editor-in-chief of Blic, Marko Stjepanović, says that AI tools are still not advanced enough to independently perform journalistic or editorial work without supervision, but they are continuously improving.

“We are becoming increasingly satisfied with AI tools”, says Stjepanović.

Blic's plan is to stay ahead of AI trends not only to sustain its operations but to improve them.

“AI tools will truly become useful as they save time—and time is money. In the near future, AI tools will be personalized for each individual journalist. These personal AI assistants will continuously evolve based on examples drawn from each journalist’s daily experiences”, Stjepanović believes.

Regulating Use

In Bosnia and Herzegovina, there is neither a legal nor a self-regulatory framework for the use of artificial intelligence in the work of journalists and media outlets.

The Press and Online Media Council in BiH says it plans to adopt guidelines for the use of AI tools in journalism in the near future.

This document will not be legally binding, but it will serve as an ethical guide for journalists.

The Communications Regulatory Agency of BiH (CRA), on the other hand, obliges licence holders for radio and TV stations to comply with existing regulations and guidelines. However, none of those currently address the use of artificial intelligence.

The CRA emphasizes that the capabilities and ways of using AI in the media are not only new for media professionals but also raise many questions for regulatory bodies—even in European countries.

“At present, there is no unified practice on this issue. In some countries, there are guidelines for the ethical use of artificial intelligence in the media, initiated by journalists’ associations as an attempt to establish basic principles for responsible AI use”, the CRA notes.

They stress that despite the many advantages of using AI, there are numerous challenges and questions that need to be addressed, such as the ethical use of data and algorithmic transparency.

“Education, regulation, and transparency are key for the responsible use of AI technologies, especially in political or electoral contexts. In that regard, it is worth mentioning the EU Artificial Intelligence Act, which came into force in EU member states in mid-2023 as an attempt to regulate AI and set clear guidelines and standards for the development and use of AI technologies”, the CRA states.

The Act sets rules for placing general-purpose AI models on the market, rules on data quality, bans on certain AI practices, transparency requirements for AI systems, and rules on market oversight and accountability across the EU.

“At this stage, it is too early to speak about specific regulatory approaches to this issue in BiH from the Agency’s perspective. However, it is important to keep in mind that, as a candidate country, Bosnia and Herzegovina is obligated to incorporate EU legislation into its national framework—including this regulation and other laws related to the use of digital services, which are already being implemented in EU countries”, the CRA concluded.

Ethical Dilemmas

While regulation and self-regulation are pending, journalists rely on personal ethical standards.

For the editor of Fokus, ethical dilemmas are the biggest issue related to the use of AI tools.

“That’s why we avoid using AI tools to their full capacity—to prevent any breach of ethical standards. With its massive database, AI can produce fake news more convincingly than a human”, says Dundić.

He believes that in the future, the greatest focus will need to be placed precisely on ethical norms.

“Education and training will be very important in the coming period to ensure that AI tools are used in an ethically acceptable way and to prevent potential misuse”, says Dundić.

For the editor-in-chief of Serbia’s most-read news portal, the use of AI tools raises ethical questions particularly in the production of sound, images, and videos.

“By the end of this decade, AI will be so advanced that, for example, during political campaigns, we won’t be able to say with certainty whether a political figure actually said or did something we see in a video—or whether it’s a perfectly realistic AI simulation of reality”, warns Stjepanović.

The tools used by journalists at Klix don’t raise major ethical concerns, as they are primarily related to the technical aspects of journalistic work.

“However, this is a topic that must be continuously examined and improved”, says the editor-in-chief of the most-read media outlet in Bosnia and Herzegovina.

Tepina says that ethical dilemmas are resolved by establishing order: transparency in usage, control over algorithms, and editorial oversight.

“No information should be published automatically without being reviewed by responsible editors”, advises Tepina.

Kulenović explains that ethical dilemmas are the greatest and most understandable barrier to the broader use of AI tools in journalism. However, that doesn’t mean that small, ethically grounded experiments can’t be done.

“What’s most important right now—and more achievable—is for media outlets to invest in human capacity that will better understand all these changes and report on them without relying too heavily on what are often unreliable sources”, adds Kulenović.

Will AI Replace Journalists?

Public discussions on whether AI can replace journalists and editors have been ongoing for some time, but Mediacentar’s interviewees believe this will not happen. The human touch, spoken word, and written language, they agree, are irreplaceable.

“I don’t believe automation will eliminate jobs, but it will certainly change them”, reflects Tepina. The editor of Spin.info believes that AI won’t be able to replace certain professions—journalism among them.

The editor at FACE TV shares a similar opinion.

“Automation through AI tools poses a greater threat to jobs in other sectors, such as customer service, call centres, and even some IT positions. When it comes to the media, AI will never be able to replace journalists or presenters, primarily for two reasons: it lacks the ‘human touch’, the unique voice and writing style each journalist brings, and AI will never be able to uncover exclusive information on its own”, says Javor Ibrahimbegović confidently.

AI will eliminate some jobs in the future but will also create new positions in emerging professions, Kulenović believes. He notes that AI may replace journalists when it comes to writing short news briefs, which would free up time for them to focus on in-depth journalistic work.

This article was produced with the financial support of the European Union. Its contents are the sole responsibility of the Mediacentar Foundation and do not necessarily reflect the views of the European Union.