US EditionAI assistants make widespread errors about the news, new research shows
A study of 3,000 AI responses across 18 countries found 45% had significant issues, including 31% with serious sourcing errors and 20% with major accuracy problems.
- On October 22, 2025, the European Broadcasting Union and BBC released a study showing 45% of AI answers from ChatGPT, Microsoft's Copilot, Google's Gemini, and Perplexity had significant issues.
- The study enlisted 22 public service media organisations across 18 countries and 14 languages, posing the same 30 news-related questions between late May and early June during publishers' two-week content access window.
- Accuracy failures—including hallucinations and outdated facts—accounted for 20% of errors, while sourcing errors affected 31% of responses and Gemini, Google's AI assistant, had issues in 76% of replies.
- The broadcasters and media organisations behind the study are calling for national governments and AI companies to act, launching the 'Facts In: Facts Out' campaign and the News Integrity in AI Assistants Toolkit.
- Only 7% of online news consumers use AI chatbots for news, rising to 15% among under-25s, and Jean Philip De Tender warned `This research conclusively shows that these failings are not isolated incidents`.
91 Articles
91 Articles
The main artificial intelligence assistants (AIs) distort the content of the news in almost half of their responses, according to a new study published on Wednesday by the European Broadcasting Union (ERU) and the BBC.The international research studied 3000 answers to questions about news from the main artificial intelligence assistants, computer tools that use AI to understand orders in natural language and complete tasks for a user.The accurac…
AI chatbots make mistakes with news content nearly half of the time, says study (Business)
A new report from a global alliance of public broadcasters says AI chatbots make mistakes with news content nearly half of the time. The study, which looked at how AI chatbots answer questions about news and current affairs, involved 22 public media organizations in 18 countries, including CBC/Radio...
IA's assistants like ChatGPT make mistakes in almost half of the time when they were asked about current issues, a study published today and coordinated by the UER.
Nearly half of all AI-generated search results for verified news contain errors, according to a survey conducted by public service media in 22 countries.
Coverage Details
Bias Distribution
- 54% of the sources lean Left
Factuality
To view factuality data please Upgrade to Premium
