Almost 10% of articles in U.S. newspapers contain some text written with artificial intelligence (AI), according to a new study by the University of Maryland.
The report, based on a tool developed by Pangram Labs, found that of 186,000 articles studied in the summer of 2025, 9.1% were either AI-generated (5.2%) or mixed (3.9%). The remaining
articles were classified as written by humans.
But this varies with the size of the publication. Among papers with circulation of more than 100,000, a mere 1.7 of
the articles were AI-generated, either partially or fully. But at smaller ones, 9.3% contained significant AI content.
In addition, the study found 219 articles
that included AI content on the opinion pages of The Washington Post, The New York Times and The Wall Street Journal.
However, many of these were written by public
figures and were 6.4 times more likely to contain AI-generated content than news articles in the same newspapers.
advertisement
advertisement
Moreover, the study found that “AI use is largely
undisclosed.”
“Disclosure of AI use (e.g., exactly how and where AI was used in the construction of the article) is especially important to maintain
audience trust,” the study states.
But it adds: “Unfortunately, in a sample of 100 AI-flagged articles from unique newspapers in the recent news dataset, we
find that 95% of authors and 91% of publishers did not disclose AI use. The few disclosures we observed appeared only in environmental reporting, such as weather forecasts and air quality
alerts.”
The researchers analyzed work by 10 veteran reporters who were active prior to ChatGPT’s release. They were evenly split between men and
women.
AI usage by this set rose from roughly zero in 2022 to 15.7% in 2023, 36.1% in 2024, and 40.4% in 2025. But no one disclosed this
use.