A recent study jointly conducted by the European Broadcasting Union (EBU) and the BBC has raised significant concerns about the reliability of leading artificial intelligence (AI) chatbots when it comes to reporting news. The investigation, which encompassed 18 countries and 14 different languages, aimed to evaluate the accuracy and trustworthiness of popular AI systems such as OpenAI’s ChatGPT, Microsoft’s Copilot, Google’s Gemini, and Perplexity. Professional journalists were tasked with assessing thousands of AI-generated responses to recent news stories, focusing on key factors including factual accuracy, the use of credible sources, and the ability to distinguish between fact and opinion.
The findings were troubling: nearly half (45%) of all analyzed chatbot responses were found to have at least one significant issue, according to a summary from the BBC. Even more concerning, about one in five responses (20%) contained major accuracy problems. These included hallucinations—where the AI invents information and presents it as verified fact—or the dissemination of outdated or misleading data. Among the four AI systems evaluated, Google’s Gemini performed the worst, with a staggering 76% of its responses exhibiting significant issues, particularly in terms of providing proper source citations.
This research comes at a crucial moment, as generative AI tools are increasingly being used as the primary interface for internet searches and news consumption. Traditionally, people have relied on established news organizations and search engines to inform them about current events. However, the rise of AI-powered chatbots is beginning to shift this dynamic, raising concerns about the potential impact on public discourse and trust in the media.
Recent surveys reflect this changing landscape. According to the Reuters Institute's Digital News Report 2025, 7% of people worldwide now use AI tools to keep up with news updates. This figure rises to 15% among those under the age of 25, indicating that younger generations are especially likely to turn to AI for news. In contrast, a Pew Research poll in the United States conducted in August showed that 75% of adults say they never get their news from AI chatbots. Despite this, the overall trend points toward increasing reliance on AI-generated information, especially among digital natives.
Adding to the complexity is the way people interact with AI-generated news. Even though few users express complete trust in information provided by features like Google’s AI Overviews (which utilizes Gemini), most rarely take the time to verify the accuracy of responses by clicking on source links. This passive consumption of potentially flawed information increases the risk that misinformation could spread unchecked, further undermining confidence in credible journalism.
The EBU and BBC warn that the persistence of these issues could have serious societal and political consequences. Jean Philip De Tender, EBU Media Director and Deputy Director General, emphasized that the study’s findings are not isolated or anecdotal. “They are systemic, cross-border, and multilingual, and we believe this endangers public trust,” he stated. If people lose the ability to discern fact from fiction, the organizations argue