Balkan Press Councils Add AI Rules to Journalism Codes
By Suad Seferi ·
Updated ethical standards across South-East Europe are setting clearer expectations for the use of artificial intelligence in newsrooms, including human verification, editorial approval and disclosure of AI-generated material. Press and media councils across South-East Europe are updating their journalism codes to address the growing use of artificial intelligence in reporting and content production. The changes are intended to provide practical guidance for newsrooms already using AI for tasks such as transcription, translation, research, image production and text generation. The latest development came from Montenegro, where a revised Code of Journalists was presented in Podgorica on July 8, 2026. The updated code states that AI-generated or AI-assisted material should not be published without human verification and editorial approval. It also says that the use of artificial intelligence should be clearly disclosed when a lack of labelling could mislead the audience. The central principle is that journalists and editors remain responsible for everything they publish, regardless of which technology was used to create or modify it. What the new rules mean The updated codes do not prohibit journalists from using artificial intelligence. Instead, they establish boundaries for how AI can be included in professional journalism. A journalist may use an AI tool to transcribe an interview, translate material or organise information. The newsroom must still verify the result before publication. The same principle applies when AI is used to generate text, images, audio or video. The output cannot be treated as accurate simply because it appears convincing or is produced by a widely used system. Human verification means that a journalist or editor must review the content, compare it with the original material where necessary and confirm that it meets the same standards expected from traditionally produced reporting. Editorial approval also means that responsibility remains with the newsroom. An error cannot be explained away by saying that it originated from an AI system. When should AI content be labelled? One of the main questions addressed by the regional standards is whether audiences should be informed when artificial intelligence played a role in producing content. The regional declaration adopted by press councils in Ohrid in May 2025 states that the public has a right to understand when and how AI meaningfully shapes the information it receives. It calls for transparency and labelling that is proportionate to the role AI played in creating, selecting or processing the content. This does not necessarily mean that every minor use of an automated tool requires a warning. Correcting spelling with software is different from publishing an entirely AI-generated illustration or an article substantially written by a language model. The relevant question is whether the use of AI could change how the audience understands the origin, authenticity or reliability of the material. An AI-generated photograph presented alongside a news report, for example, should not be allowed to appear as documentation of a real event. Synthetic audio, video or quotations also require clear treatment because audiences may otherwise assume they are authentic. A broader regional process Montenegro’s revised code is part of a wider effort across South-East Europe. The Serbian Press Council adopted a revised journalists’ code in December 2024. Albania amended its Code of Ethics for Journalists in 2025, while the Turkish Press Council updated its professional principles during the same year. North Macedonia presented a new journalism code in Skopje in mid-2026. Bosnia and Herzegovina is also revising its existing code, with adoption planned by the end of 2026. Its update is expected to address deepfakes, algorithmic bias and automated content generation. The national codes are not identical, but they follow a shared regional position developed by press and media councils from South-East Europe and Türkiye. At their meeting in Ohrid in May 2025, the councils agreed that AI should support journalism rather than replace human judgement, professional responsibility or editorial control. The role of AI in a newsroom Artificial intelligence can be useful in journalism when it assists with repetitive or time-consuming work. It can turn recordings into transcripts, search large documents, translate interviews, create summaries and help journalists identify information that requires further investigation. However, AI systems can also produce incorrect facts, invented sources, inaccurate translations and realistic synthetic material. This is why the new ethical rules focus less on the tool itself and more on the process surrounding its use. Before publishing AI-assisted material, newsrooms are expected to ask basic editorial questions: Was the information checked against reliable sources? Was the original recording, document or image reviewed? Could the audience misunderstand how the content was produced? Is a clear label necessary? Who approved the final publication? These are not entirely new questions. They are traditional journalism standards applied to technologies that can generate material more quickly and at a larger scale. Responsibility remains with journalists The regional codes make clear that artificial intelligence is not an author, editor or accountable media organisation. It cannot defend a reporting decision, correct an error independently or accept responsibility for misleading the public. That responsibility continues to belong to journalists, editors and publishers. The updated codes therefore do not attempt to create a separate form of journalism for the AI era. They extend existing principles such as accuracy, transparency, fairness and editorial accountability to new tools. The practical message for newsrooms is simple: AI may help produce journalism, but it cannot be the final authority on what gets published. A…