A new investigation has found that the most popular news app in the United States published more than 30 articles in the past three years that were inaccurate, altered or signed by AI, with real-world consequences.
NewsBreak, the most popular news app in the US, markets itself as a local news source, ranking number one on the Google Play Store with over 50 million downloads, and topping the Apple App Store news charts behind only X and Reddit.
The app operates only in the US and acts as an aggregator that brings together news from various outlets such as Fox, Reuters, and CNN on one platform.
Reuters reported on Wednesday that Newsbreak has used AI at least 40 times since 2021 to publish inaccurate articles, post articles from other sources with fake bylines or grab content from competitors.
For example, two AI-based articles in NewsBreak incorrectly stated that Pennsylvania-based charity Harvest912 was hosting 24-hour health clinics for the homeless.
“By putting out this false information, homeless people will be encouraged to walk to these venues to attend clinics that are not actually taking place,” Harvest 912 said in an email to News Break in January.
Another email sent to NewsBreak from Food to Power, a Colorado-based food bank, detailed how NewsBreak had misrepresented the timing of three separate food distributions in January, February and March.
Food banks have had to explain the problem to people who had turned up in response to the News Break article, and send them home without the food they were expecting.
Newsbreak told Reuters it had removed five articles containing inaccurate information.
Related: Microsoft replaced news editors with AI, and it's leading to one disaster after another
Regarding the AI tools and fake bylines, NewsBreak appears to have used five pseudonyms as bylines when reposting AI-generated articles from other sites.
Former Newsbreak consultant and former Wall Street Journal editor-in-chief Noam Pearlstein addressed the issue in an internal memo to Newsbreak CEO Jeff Jen in May 2022, writing, “I can't think of a faster way to destroy the Newsbreak brand.”
Zheng responded to the memo, acknowledging the problem and asking his team to fix it.
Related: OpenAI gives users access to Financial Times articles to train their AI
NewsBreak isn't the only news organization under scrutiny for its AI content: Bloomberg reported earlier this month that local San Francisco newspaper Hoodline was relying on AI to churn out stories, and at one point attributed those stories to unique AI personas with their own bios.
AI has also been known to generate inaccurate content: News outlet CNET used AI to create more than 70 articles last year, but had to correct many of them due to factual errors.
Meanwhile, Google announced “more than a dozen technical improvements” last week after users discovered that its search engine's AI-generated overviews were giving some inaccurate answers.
Related: Google's AI overview is already getting it wrong
