Japan Ranks 3rd in the World for Accesses to AI Deepfake Sexual Image Sites

A survey by the Yomiuri Shimbun has found that the number of visits from within Japan to a website that uses generative AI (artificial intelligence) to create “deep fake” sexual images rose to more than 18 million times in the 12 months to November, the third highest in the world after the United States and India. The number of cases of fake sexual images of real people being created and spread on social media is increasing both at home and abroad, and these sites are believed to be breeding grounds for this. Experts also call for the introduction of legal regulations in Japan to prevent damage. The Yomiuri Shimbun identified 41 websites claiming to be able to create fake sexual images. Using a tool from site analysis company SimilarWeb, we looked at the number of hits by country for 41 sites from December last year to November this year. As a result, the United States had the most with approximately 59.73 million doses, followed by India with approximately 24.57 million doses, Japan with approximately 18.43 million doses, Russia with approximately 17.59 million doses, and Germany with approximately 16.86 million doses. The average number of monthly visitors in Japan was approximately 410,000, 80% of which were from smartphones. These sites allow users to upload image data of real people and have it edited to make it appear as if they are naked. Instructions for use are mainly in English and Russian, but there is also a Japanese website. More than half of them are believed to have been released this year. As the number of such sites increases, there is an increasing number of cases where fake images and videos are created and distributed without permission. According to a survey by U.S. security company Security Hero, the number of deepfake videos found online in 2023 was 95,820, 5.5 times the number in 2019, and 98% of them were sexual videos. In response to the growing damage, there are growing moves towards legal regulation overseas. Professor Ichiro Sato (information studies) of the National Institute of Informatics said, “To prevent people from easily creating fake sexual images, Japan also needs to consider introducing legal restrictions and provide information literacy education to prevent perpetrators from being created.” [Yomiuri Shimbun]
2024/12/29 05:00.

ＡＩによる性的な偽画像作成サイトへのアクセス、世界３番目１８００万回…被害拡散の温床

【読売新聞】　生成ＡＩ（人工知能）で性的な偽画像「ディープフェイク」を作成できるサイトへの日本国内からのアクセス数が、１１月までの１年間で１８００万回超に上り、米国、インドに続き世界で３番目に多かったことが読売新聞の調査でわかった。

www.yomiuri.co.jp