Artificial intelligence is used by more and more journalistic organizations in the publishing process and to support the editorial work. AI offers Uitgevers opportunities, but there are also risks lurking. An overview.
BBC News: Personalization
BBC News uses AI to offer personalized content to users. The new department, BBC News Growth, Innovation and AI, mainly focuses on a young target group of news consumers. AI is used to select stories for users on their phones based on their measured behavior and preferences. That is similar to the way in which social media offer content to users.
Deborah Turness, the Chief Executive of BBC News, says that the organization “Ruthlesly focused” should be on understanding the needs of the public and delivering content in the places where users want to consume. AI is going to help With personalizations And recommendations to realize innovation and growth.
New York Times: editorial support
The New York Times introduces ‘Echo’, an AI tool that supports journalists in their work. Get editors via Echo Summaries of their articles, texts for social media and suggestions for better heads. Employees are trained to work with AI. Not only to make articles better, but also to develop news quizzes, FAQs and prepare interviews. There are limitations: Journalists from The New York Times are not allowed to use AI to write an article or change considerably or by publishing AI generated images or videos.
Le Monde: Translations
The publisher of Le Monde has been bilingual since 2023. The French newspaper has launched an English version to attract larger groups of readers. The idea of launching ‘Le Monde in English’ is not new. 50 years ago a physical English -language newspaper was published. Success on the international market only recently became feasible, thanks to digitization. The digital newspaper made global distribution a lot easier and more cost -efficient.
Certainly through the deployment from AI translations. According to Arnaud Aubron, Head of Development at Le Monde, AI is used to translate 30 Long Reads every day. He emphasizes: “That the quality comes first.” The first translation is made by the AI tool ‘Deepl’. Then a human translator will refine the text. The last step is to read by a journalist, to make sure that the article matches the tone and quality of Le Monde.
Initially, Le Monde thought that one human revision round after AI translation would be sufficient, but soon a second check turned out to be necessary to guarantee the high linguistic standard. Le Monde Deepl is also adapted to the tone and style guide of Le Monde.
Bild: Chat and ask questions
The Business Innovation team of Axel Springer, publisher of Bild, has launched an AI assistant with the name ‘Hey_’. This chatbot has been developed based on various GPT models. Hey_ is a new digital tool for Bild that aims to make journalism more interactive and more personal. Of Hey_ Users can immediately discuss articles. This makes the content of Bild interactive and dynamic.
The chatbot converts news articles into conversations and gives Bild readers an interactive and personalized experience. The AI offers both pre -set answer options and the possibility to enter free text. As a result, Hey_ constantly fits in with the questions and needs of the user.
According to Valentin Schöndienst, SVP Business Innovation at Axel Springer, Hey_ offers an accessible way to use AI for information, help and entertainment. Axel Springer wants to become a starting point for millions of Germans to get to know and use AI.
Los Angeles Times: alternative views
The Los Angeles Times experimented with a new AI tool called ‘Insights’. The AI tool aims to offer alternative perspectives with articles with a certain point of view, such as opinion. De La Times wants to help readers to understand where the published opinions are within the political spectrum, and to accompany them from various (summarized) views. The tool was powered by the LLM search engine Perplexity and used Partician to determine the political direction. It is not a great success yet. The tool was removed from an article again after a day after it seemed to be the history of the KKK. This incident led to concern for the trade union of journalists of De La Times, the uncontrolled AI-generated analyzes could harm confidence in the media.
Apple: Notifications with summaries of the news
Apple uses AI to summarize news items from large news brands for notifications on iPhones. And that is not going well. Various organizations, such as the BBC, have submitted their complaints to Apple about the inaccuracies in the AI-generated news summaries. In December 2024, the BBC filed a complaint with Apple after an AI summary wrongly claimed that Luigi Mangione, the suspect of the murder of UnitedHealthcare CEO Brian Thompson, had committed suicide.
Apple initially did not respond to this complaint, but eventually acknowledged in January 2025 that there were problems with the AI function. Since then more mistakes have come to light. This has led to increasing pressure on Apple to improve or lift the AI function. Journalistic organizations, Like the BBChave expressed concern about the distribution of disinformation and undermining trust in news sources. The BBC did a test itself and let the four large AI players, Chatgpt, Perplexity, Co Pilot and Gemini summarize all news items for a period. The researchers found errors in the summaries in half of the cases.
In response to the fuss, Apple finally decided to eliminate the AI function for news notifications for the time being and promises to make improvements in future software updates.
Conclusion
The research of the BBC and the experiment of De La Times shows the potential risks and challenges of the use of AI for generating editorial content or analysis of news items. And they count as a valuable warning for other publishers. On the other hand, there are fortunately many journalistic organizations that do not be put off too much by the pitfalls of AI. The complexity of working with Large Language Models is great. Opportunities weigh just as heavy as risks. The advice for publishers is to get inspiration in the above examples and to start the development of AI tools. But, a warned person counts for two: always leave the output of AI tools controlls and use AI mainly in co-production with human editors.
Team Bladendokter regularly organizes the training: AI on the editors. Training AI on the editors. During this training, participants gain insight into the opportunities and pitfalls of using AI in editorial practice.
With then on: Emiliano Vittoriosi For the photo via Unsplash