“AI assistants… routinely misrepresent news content no matter which language, territory, or AI platform is tested.” Damn. That’s the introduction by the BBC’s Media Centre for a study coordinated by the European Broadcasting Union (EBU) and led by the British Broadcasting Corporation (BBC), a UK-government-owned media organization.
That’s bad, bad news for the 7 percent of us who reportedly use AI to get our news, including 15 percent of people under 25 years old, according to the Reuters Institute’s Digital News Report 2025.
The study looked at how accurately the most popular generative AIs, including OpenAI ChatGPT, Google Gemini, Microsoft Copilot, and Perplexity AI, could recite news stories to users. Notably, it didn’t examine Anthropic Claude or xAI Grok.
Take a look at some of the key takeaways from the study, which was released on Wednesday, October 22, 2025.
The results are in: AI Can’t Summarize News
Overall, when the responses of all AIs were grouped together, significant issues were found in 45 percent of all answers. That’s, frankly, god awful. Thirty-one percent of responses “showed serious sourcing problems, (such as) missing, misleading, or incorrect attributions,” the study concluded, and 20 percent “contained major accuracy issues, including hallucinated details, and outdated information.”
Google Gemini was the worst of the bunch, and not by a little. The study’s authors found what they called “significant issues” in 76 percent of its responses, more than twice as many as any of the other AIs tested. The authors attributed it to poor sourcing performance.
“This study (was) built on research by the BBC published in February 2025, which first highlighted AI’s problems in handling news,” the authors wrote. “This second round expanded the scope internationally, confirming that the issue is systemic and is not tied to language, market or AI assistant.”
At least it’s getting better, right? Right? “Comparison between the BBC’s results earlier this year and this study show some improvements but still high levels of errors,” the study said.
Ah, well. Maybe you should just get your news straight from the source instead of through AI.
The post Extra, Extra, Read All About It: AI Sucks at Summarizing the News appeared first on VICE.
