The reality of AI in publishing: From testing to daily newsroom integration

Media organisations have spent the last two years experimenting with artificial intelligence, but making it a core part of the daily newsroom workflow remains a hurdle. At the FIPP World Media Congress, NewsLabs showcased how purpose-built enterprise tools are changing the way journalists handle routine reporting. A new study also reveals that readers are largely indifferent to AI transparency labels.

When generative AI first emerged, journalists feared for their jobs, but the reality has proven to be quite different. Ivan Šarić, co-founder and chief executive of NewsLabs, told delegates at the FIPP World Media Congress that artificial intelligence is not replacing reporters. Instead, it is handling the heavy lifting of repetitive tasks, allowing editorial teams to focus on authentic and investigative work. NewsLabs started with three people two years ago and has expanded to a team of 14, developing a platform tailored specifically for publishers in Central and Eastern Europe. Media companies are finding that without a dedicated AI solution, they quickly fall behind competitors who can now produce routine news at a fraction of the traditional time.

Transparency labels do not damage reader trust

One of the biggest concerns for publishers regarding AI integration has been the potential loss of credibility if readers know a machine wrote the text. To test this, NewsLabs conducted a scientific study involving 655 readers across seven countries, including the United Kingdom, United States, France, Germany, Italy, Spain, and Croatia. Readers were shown articles with three different bylines: written entirely by a journalist, written by a journalist with AI assistance, and generated entirely by AI. The results were surprising, showing that a transparency label did not change how credible the article looked to the audience. In fact, more than 60 percent of readers did not even remember seeing a label after finishing a few articles.

Furthermore, readers were just as willing to read the AI-assisted articles to the end as they were the purely human-written ones. The study concluded that for newsrooms, transparency about AI involvement does not negatively impact audience trust. Šarić noted that while the European Union has introduced regulations requiring labels for AI-drafted content that lacks human oversight, publishers should not fear a backlash. The data clearly indicates that readers are largely indifferent to the use of artificial intelligence, provided the quality of the journalism remains high and the content is relevant to their interests.

Automating routine tasks and data analysis

The practical application of enterprise AI solutions goes far beyond basic text generation. During a live demonstration, the NewsLabs platform showed its ability to automatically generate news articles from social media posts on platforms like Instagram and Reddit. It can also ingest hours of video, such as a lengthy parliamentary session, and transcribe, summarise, and draft relevant news stories in just a few minutes. This feature has been particularly useful for journalists who previously had to sit through eight-hour meetings just to extract two viable stories. By automating this process, reporters are freed up to make calls, conduct interviews, and develop original angles.

Another significant advantage demonstrated was the rapid processing of complex, structured data. Šarić highlighted how the system can automatically download and analyse extensive spreadsheet data, such as monthly tourism statistics from the Croatian Bureau of Statistics. What would normally take an economics reporter hours to digest can be processed by the AI to produce a comprehensive, accurate news draft almost instantly. The platform ensures that the output matches the specific tone and style of the individual media brand, allowing editors to retain full control while significantly optimising their daily workflow.

As media houses move from merely testing consumer tools to deploying tailored enterprise solutions, the emphasis is shifting towards data security and workflow efficiency. With AI handling the repetitive content of daily news, journalists are finally getting the time to pursue the original stories that truly matter.