New research led by the BBC and coordinated by the European Broadcasting Union (EBU) has revealed that AI assistants consistently misrepresent news content across various languages and regions. The extensive study, launched at the EBU News Assembly, involved 22 public service media organizations across 18 countries and assessed over 3,000 responses generated by four leading AI tools: ChatGPT, Copilot, Gemini, and Perplexity.
The research uncovers substantial systemic issues, with 45% of AI-generated answers exhibiting significant problems. Notably, 31% of the responses suffered from serious sourcing deficiencies, including missing or incorrect attributions, while 20% contained major accuracy issues such as outdated information and hallucinations. Gemini was identified as the least reliable tool, demonstrating significant issues in 76% of its responses.
This misrepresentation is critical, as AI assistants are increasingly replacing traditional search engines for news consumption. According to the Reuters Institute’s Digital News Report 2025, 7% of online news consumers rely on AI for their news, a figure rising to 15% among individuals under 25. EBU Media Director Jean Philip De Tender emphasized that these findings indicate systemic problems that undermine public trust in news sources.
Peter Archer, BBC Programme Director for Generative AI, acknowledged the excitement surrounding AI but highlighted the necessity for trust in the information provided. He noted that improvements in AI tools remain insufficient to address significant concerns.
The research team has released a News Integrity in AI Assistants Toolkit aimed at developing solutions to the identified challenges, focusing on improving response accuracy and enhancing media literacy among users. Additionally, the EBU and its members are calling on EU and national regulators to enforce existing laws on information integrity and digital services while advocating for ongoing independent monitoring of AI tools.
This study builds upon earlier research published by the BBC in February 2025, confirming that these issues persist across different languages and markets. Participating broadcasters included organizations from countries such as Canada, Germany, Italy, and the United Kingdom.
In a separate report, the BBC has indicated that many UK adults trust AI to produce accurate news summaries, with one-third expressing confidence in AI accuracy—this figure rises to nearly half among those under 35. Concerns were raised about the public perception of AI news summaries and the potential for misplaced blame on news providers for inaccuracies generated by AI, which could negatively impact trust in journalistic entities.
