Artificial intelligence is also becoming an increasingly present tool in investigative journalism. Technological progress enables faster processing of large amounts of information, which can be of particular importance in investigations into corruption and abuse of power.

 

In Bosnia and Herzegovina, there are no publicly available and documented examples in which a specific AI tool has directly contributed to an anti-corruption investigation. According to a study presented at a conference on the use of artificial intelligence in the media in BiH last year, organized by Burch University, two-thirds of surveyed participants (62%) reported using some form of AI in their daily work. Nonetheless, they face significant challenges related to the knowledge required for effective and ethical application of these tools. Most commonly, AI is used for searches, translations, and basic data processing, with no confirmed examples of its application in uncovering corruption networks.

 

Investigative media such as BIRN and the Center for Investigative Reporting (CIN) point to the potential of AI, particularly in the analysis of public procurement. For now, however, such methods remain more of a theoretical recommendation than practically implemented tools. Still, these outlets do make use of various tools, including those supported by AI, especially for transcription or, in the case of CIN, for publishing content on social media.

Secondary sources, such as the Reuters Institute report from 2023, indicate that AI is most often used in the search and data organization phases, while its use in actual writing remains a cautious due to concerns about accuracy.

 

 

Various tools, including those powered by artificial intelligence, now help journalists analyse financial flows faster, search databases, detect irregularities, and link different sources of information. Programs such as Aleph, Hunchly, Google Pinpoint, and OCR tools are used for the automatic collection, recognition, and organization of data.

 

Semir Mujkić, editor of BIRN BiH, notes that although AI tools are available, their newsroom uses makes only limited use of them.

 

“At BIRN, we only use basic tools that fall under artificial intelligence, such as ChatGPT—and even then, only for simple queries. However, when it comes to serious analyses and investigative work, we use artificial intelligence very sparingly, primarily due to ethical dilemmas and the unreliability of these tools. They often fail to provide accurate or complete information, which is why we cannot treat them as a primary source,” says Mujkić.

 

The most common dilemmas concern accuracy (AI sometimes “invents” information), copyright (particularly in relation to generated content), and transparency in work, since it is often unclear which data AI draws from when producing answers. The use of AI in journalism also raises questions of accountability: if the algorithm makes a mistake, who bears the consequences—the journalist, the editor, or the technology itself?

 

On the other hand, Mujkić points out that there have been cases where AI was used in specific situations, such as analysing Google satellite images to detect illegal logging, but these remain only early attempts.

There are, however, useful tools and platforms that can be support investigative journalism in analysing satellite images, particularly for topics such as environmental degradation, biodiversity loss, illegal construction, deforestation, and mining.

 

For instance, the Global Forest Watch – University of Maryland team used Google Earth Engine, a cloud-based platform for satellite data and analysis, to track global changes in forest cover over several years. By examining Landsat satellite images, the platform identified locations where forests had disappeared, with an accuracy of up to 30 metres, and presented the results as interactive maps that were later published by other media.

 

In addition to this tool, platforms such as Sentinel Hub for analysing water surfaces, fires, and similar phenomena, Microsoft AI for Earth for monitoring soil degradation and deforestation, and SkyWatch EarthCache for investigating illegal construction demonstrate the potential of AI-supported satellite analysis in investigative reporting.

 

 

Adnan Čomor, editor of digital media at the Center for Investigative Reporting, emphasizes that working in a small team requires making the most of all available resources, including AI-based tools.

 

“We work in a small team and have to use all the tools at our disposal. AI is unavoidable today and makes many processes easier, especially when it comes to working on digital platforms,” says Čomor.

 

He adds that several tools significantly help him and his colleagues in CIN’s digital department in their daily work.

“I use the tool Kapwing, which helps me in several ways—from video transcription and cropping material to audiovisual support. I also use Adobe Express very often, since in just a few short steps I can adapt any story segment to the level required by social networks,” says Čomor.

 

Alongside tools such as Aleph, it is also worth mentioning Pandas and OpenRefine, often used by foreign media for processing structured data. Pandas allows researchers to efficiently analyse and transform tabular data, while OpenRefine supports “data cleaning”—merging duplicates, standardizing formats, and improving interpretation of large datasets.

Nicholas Diakopoulos, one of the world’s leading experts on the use of AI in journalism and a professor of communication studies and computer science at Northwestern University, believes that artificial intelligence should not replace journalists but rather support them in data analysis, the automation of routine tasks, and the detection of patterns in large databases.

 

In his book Automating the News: How Algorithms Are Rewriting the Media, Diakopoulos warns of the dangers of uncritical adoption of algorithms in newsrooms, particularly in relation to bias, lack of transparency, and content personalization. He notes that algorithms are already changing how news is produced and circulated, especially through automated news writing and content personalization. For example, Heliograf, an automated system used by the Washington Post to report on the Olympic Games and US elections, produced hundreds of short news items from structured data without human intervention, enabling the newsroom to cover more local electoral units and sporting events than would have been possible with journalists alone. Similarly, Bloomberg uses automation in financial journalism to report on market changes in real time.

 

However, Diakopoulos warns that with automation must undermine the basic principles of journalism—accuracy, accountability, and transparency–. He argues that algorithms shaping news should be understandable and publicly explained.

 

“If an algorithm decides what we will read, then we have the right to know the criteria it uses,” he says.

 

 

Mujkić, emphasizing the dedicated work of journalists, points out that there are serious problems related to the very nature of journalistic work, which cannot easily be reduced to algorithmic processes.

 

“As an editor, I have noticed that artificial intelligence changes the journalist working in the newsroom. I don’t want real people, journalists, to be replaced by a tool. I believe that it is always the journalist who should do this job and that they will always achieve better results. Those results may be more expensive and take longer, but they will certainly always be better,” he explains.

 

BIRN BiH journalists have tested how useful artificial intelligence could be for their work—for instance, in finding people mentioned in Hague Tribunal verdicts who had not been prosecuted. They discovered that AI provided completely false results, inventing facts or indictments that did not exist. This experiment highlighted the serious limitations of AI tools in working with specific legal databases and topics related to transitional justice.

 

Mujkić concludes by underlining the need to balance the use of advanced tools with preserving the essence of journalism: “I welcome every advancement in tools; we use them a lot because they make our lives easier, allowing us to do things we could never do otherwise—without satellite images, without access to databases, without data processing. We just need to be aware of their advantages and disadvantages and use them responsibly.”