Study Finds AI Models Misrepresent News Nearly Half the Time
On October 22 2025, a study published by the European Broadcasting Union and the BBC found that AI assistants mis-represented news events in about 45 % of sampled responses The research evaluated over 2,700 answers from models such as ChatGPT and Gemini, finding frequent issues in sourcing and factual accuracy. For users, this reinforces the need for caution and human verification when relying on AI-produced summaries. The significance: underscores that despite rapid advancement, AI’s reliability in journalistic or informational contexts remains imperfect.