October 14, 2025
TOKYO – Ninety-one percent of respondents to a recent Yomiuri Shimbun survey were concerned that the spread of misinformation and disinformation on social media could influence election results.
Only 7% said they were not concerned.
The nationwide survey on the media was conducted ahead of Newspaper Week, which begins on Wednesday.
Concerns were raised by 91% of respondents about misinformation and disinformation spreading on social media through video clips that show arbitrarily edited versions of candidates’ speeches and debates.
Eighty-two percent of respondents said they had high expectations for initiatives by newspaper publishers, broadcasters and other news outlets to fact-check election-related information on social media, significantly exceeding the 15% who did not.
When asked about the issue of generative artificial intelligence such as ChatGPT being trained on news articles and other copyrighted works without permission, 83% of respondents believed that rules and regulations for AI service providers should be strengthened, while only 15% said they did not believe so.
There has been a series of lawsuits against AI service providers in Europe, the United States and Japan regarding generative AI’s use of unauthorized copyrighted works.
The Japan Newspaper Publishers & Editors Association has issued a statement urging AI service providers to obtain permission from news media organizations when training generative AI on their articles.
The survey found that 70% of respondents considered newspaper reporting “reliable” — unchanged from last year’s survey — while 73% said newspapers would remain “necessary” — down from 74% in the previous survey — maintaining high percentages in both cases.
The mail-in survey was conducted from Aug. 25 to Sept. 30, covering 3,000 eligible voters nationwide. Of them, 2,039 individuals, or 68%, gave valid answers.