International research on information integrity in AI assistants reveals want for safeguards and improved accuracy : NPR Further : NPR

NPR was one among 22 public service media (PSM) organizations throughout 18 nations taking part in research, led by the BBC and European Broadcasting Union (EBU) 

At NPR, we acknowledge our duty in understanding AI’s affect on journalism and in advocating for finest practices that guarantee our reporting is represented precisely. To that finish, NPR participated in a worldwide analysis research led by the BBC and the European Broadcasting Union (EBU) on information integrity in AI assistants. This was one of many largest evaluations of its form so far, together with 22 public service media organizations throughout 18 nations, and in 14 languages.

The research’s results, launched at present by BBC and the EBU, discovered that AI assistants routinely misrepresent information content material irrespective of which language, territory, or AI platform is examined. An accompanying toolkit outlines issues that have to be solved to deal with the research’s findings.

Why did NPR take part? 

As a public media group, NPR is dedicated to delivering trusted, correct journalism to our audiences, at the same time as information consumption habits change. AI assistants are already changing serps for a lot of customers. According to the Reuters Institute’s Digital News Report 2025, 7% of whole on-line information customers use AI assistants to get their information, rising to fifteen% of under-25s. This research offered us with a novel alternative to collaborate with a well-respected set of journalism organizations to research the affect of AI summarization and illustration of reports content material.

How did NPR take part?

Fourteen members of NPR’s editorial workers volunteered to function reviewers of the AI assistants’ solutions. As a part of this research, we quickly stopped blocking related bots from accessing our content material for about two weeks to gather the required responses for our evaluation. Content blocking was then re-enabled.

What did the research discover? 

The research recognized a number of systemic points throughout 4 main AI instruments. Based on information from 18 nations and 14 languages, 45% of all AI solutions had at the very least one important difficulty, and 31% of responses confirmed severe sourcing issues — lacking, deceptive, or incorrect attributions. The full research will be discovered here.

How will NPR use this info?

The outcomes assist us take into account what safeguards and viewers schooling could also be crucial, and might inform our methods and coaching for AI adoption internally. These findings additionally reinforce the significance of our current ideas and requirements, which demand that every one last work merchandise have to be reviewed, fact-checked, and edited by people, and can’t depend on AI for accuracy. NPR additionally contributed to the News Integrity in AI Assistants Toolkit, supposed to be a useful resource for expertise firms, media organizations, the analysis neighborhood and most of the people.