Some of the world’s most widely used AI chatbots often produce inaccurate summaries of news stories. Researchers tested several popular AI assistants by asking them questions about current events and requesting summaries of articles. The results showed that many of the responses included errors, distortions, or missing context, raising concerns about the reliability of AI tools for news consumption.
The research examined chatbots such as ChatGPT, Microsoft Copilot, Google Gemini, and Perplexity AI. Journalists asked these systems 100 questions using BBC articles as sources. According to the findings, about 51% of the answers contained significant problems, while roughly 19% included factual errors such as incorrect dates, statistics, or statements.
Some of the mistakes were striking. For example, one chatbot incorrectly claimed that the UK’s National Health Service did not recommend vaping as a method to quit smoking, while others wrongly stated that certain political leaders were still in office after leaving their positions. These errors often occurred because AI systems mixed facts with outdated information or failed to clearly separate opinion from verified reporting.
Media organizations warn that such inaccuracies could undermine public trust in reliable journalism. The BBC’s leadership cautioned that AI tools offer great opportunities but could also spread misinformation if not carefully managed. Experts are urging closer cooperation between technology companies and news organizations to improve accuracy, transparency, and attribution when AI systems summarize or reference news content.
