Ahead of the 14th SoMo Borac, the regional competition for the best digital projects taking place on 23 October, co-organized for the fourth consecutive year by the Croatian Digital Publishers Association (HUDI), we are opening a question that increasingly occupies newsrooms: how can the credibility of content be preserved when the path from source to audience runs through platforms, algorithms and AI assistants?
The column is written by Assoc. Prof. Marija Slijepčević, PhD from VERN’ University, President of HUDI’s Content Integrity Team, which brings together representatives of leading Croatian digital publishers and the academic community.
Almost a decade after Donald Trump turned the term fake news into a political weapon against the media, the battle they are now fighting has become far more complex. Today, the problem is not only fabricated news. Authentic photographs, recordings and data whose context has been changed or whose source has been lost can appear equally convincing. In today’s media ecosystem, it is difficult to determine who produced the content, how it was altered, where the data, photographs and video content came from and who is responsible for them. Although we have never had access to a greater quantity of news, sources and tools for verifying them, trust in the media continues to decline and is now at its lowest level since it began being measured in 2015. According to the Reuters Institute Digital News Report 2026, across the 48 markets covered, only 37% of respondents on average say they trust most news most of the time, while concern about false and misleading content has risen to 62%. These figures point to an alarming issue of media content integrity.
What do we mean by content integrity
The term integrity comes from the Latin integritas and denotes wholeness, intactness and correctness. In media content, this means that we can determine where information comes from, whether it has been altered, the context in which it originated and who is responsible for it. This is why even an authentic recording can lose its integrity if it is published with an incorrect description or presented as a record of another event. Integrity is not guaranteed simply because content was produced by a human, just as it is not automatically compromised by every use of artificial intelligence.
In the international context, the term information integrity (information integrity) is also used. The UN associates it with the accuracy, consistency and reliability of information, while the OECD further expands the concept to include quality, consistency, clear provenance and reliability of information. Content integrity means that information has not lost what makes it credible on its journey from source to audience: accuracy, context, provenance and accountability. That journey has become longer today. Between the source and the audience, there are no longer only journalists and editors, but also platforms, algorithms, aggregators and AI assistants.
The way audiences access news has changed
The Reuters Digital News Report for 2026 records a historic shift: for the first time globally, social media and video platforms have become a more widely used route to online news than the websites and apps of media organizations. They are used for news by 54% of respondents, compared with 51% who use media websites and apps.
Across the 48 markets covered by the research, 10% of respondents say they use AI chatbots for news, compared with 7% a year earlier, while among those under 35 the share is 16%. Only 1% cite AI as their main source of news, which means it is a supplementary but rapidly growing channel.
Systematic disruption of context and sources
This mediation and simplification of media content creates numerous challenges. In an international EBU study, 45% of analyzed responses by AI assistants to questions about news contained at least one significant issue. Serious sourcing problems were found in 31% of responses, while factual accuracy problems appeared in 20%. The study involved 22 public service media organizations from 18 countries and 14 languages and analyzed more than 3,000 responses from ChatGPT, Copilot, Gemini and Perplexity.
Audiences want a human being who is accountable for what is published
Audiences clearly distinguish technological efficiency from journalistic credibility. In 2025, the Reuters Institute surveyed around 12,000 people in six countries and found that only 12% would accept news produced entirely by artificial intelligence, while 62% would accept news produced entirely by journalists. This could be a good signal to publishers that human accountability still carries reputational and market value.
Provenance is not the same as truth
Part of the response to new challenges comes from technologies for verifying content provenance. C2PA is an open technical standard on which so-called Content Credentials are based, a type of digital record that can accompany a photograph, video or other content and contain information about how it was created, which tools were used to alter it and whether artificial intelligence was used in its creation. Such a record makes it easier to verify provenance, which tools were used to alter it and whether artificial intelligence was used. Such a record makes it easier to verify provenance, but it does not guarantee truthfulness: an authentic photograph can still be published with an incorrect description or taken out of context. Data can also be lost when content is republished or captured in a screenshot.
Regulation sets the minimum, but cannot build trust on its own
The issue of content integrity is also becoming increasingly present within the regulatory framework. Croatia’s Act on the Implementation of the European Media Freedom Act requires providers of media services, electronic publications and video-sharing platforms to label content that has been “generated by artificial intelligence in any way.” This wording has raised the question of whether text should also be labeled when AI has been used only for proofreading, translation, transcription or suggesting a headline. The Croatian Digital Publishers Association (HUDI) therefore requested an interpretation from the Ministry of Culture and Media.
The Ministry distinguishes between content created by AI and content for which AI served as an auxiliary tool in the production process. Newsrooms are nevertheless left with a practical question: where does technical assistance end and content creation begin? The answer will require shared guidelines, exchange of experience and education, including through the work of HUDI’s Content Integrity Team.
At the European level, the Code of Conduct on Disinformation has been integrated into the regulatory framework of the Digital Services Act since July 2025. Its commitments serve as a relevant benchmark for assessing how large platforms and search engines that are signatories manage disinformation risks. Among other things, the Code covers transparency in political advertising, demonetization of disinformation, cooperation with fact-checkers and enabling researchers’ access to data.
Such regulations are important because they define minimum obligations and distribute responsibility more clearly. However, a legal label alone cannot confirm that content is accurate, fairly contextualized and editorially responsible. Content integrity therefore cannot be reduced to formal regulatory compliance. It is built every day through editorial decisions, source verification, transparent corrections and clear accountability for what is published.
What we are already doing in Croatia
The Reuters Institute for Croatia in 2026 states that 72% of Croatian respondents believe media owners influence reporting, while 74% say the same about governments and politicians. Content integrity is not only a question of AI and deepfakes. In the Croatian context, it particularly includes transparency of ownership, separation of editorial and commercial content, labeling of native content and sponsorships, clear editorial policies on the use of AI and protection of original journalistic content from incorrect attribution and unauthorized reuse.
Content integrity is therefore being considered increasingly seriously in Croatian newsrooms. HUDI has established the Content Integrity Team, which brings together representatives of leading Croatian digital publishers and the academic community. Its task is to preserve the credibility of content, analyze the impact of artificial intelligence, combat disinformation and develop practices for responsible content creation in the digital market. The fact that representatives of different and mutually competing media organizations sit at the same table is important in itself: content integrity is not a competitive advantage that one publisher can build in isolation, but a shared standard on which trust in the entire digital media market depends.
Responsibility is shared
The conclusions of theoretical discussions often point toward media literacy, but it is not fair to shift responsibility onto audiences who today must be able to recognize disinformation and question both sources and the motives behind the publication of content. Responsibility is certainly shared among media consumers, platforms and publishers. Newsrooms and publishers should therefore adopt public policies on the use of generative artificial intelligence, ensure mandatory human verification before publication, document sources and content creation processes, and distinguish more fairly between journalistic, sponsored and automated content. Platforms and AI systems can more clearly ensure attribution of media content through links to the original content, place greater emphasis on preserving context and the traceability of content provenance, and clearly label manipulated and synthetic content.
Trust in the media is not an abstract value. For digital publishers, it is becoming a differentiating market value. In a world where content can be produced at almost no cost, content to which someone attaches their name, reputation and responsibility will increasingly carry value. Content integrity can only be the outcome of a professional culture that the media industry still needs to build together.
