TOKYO: Over 43,000 images and videos are suspected to have used artificial intelligence (AI) to generate the likenesses or voices of Japanese celebrities and voice actors without their consent over a roughly two-month period, Kyodo News reported, citing a study by a nonprofit rights protection organisation.
The Japan Publicity Rights Protection Organisation said the posts on social media platforms - the vast majority of which were on TikTok - were viewed around 335 million times. The resulting economic losses suffered by celebrities and other rights holders are estimated at between two billion yen (US$12 million) and 4.5 billion yen.
"The widespread scale of posts suspected of unauthorised use is deeply concerning. As it is particularly difficult to determine whether a voice matches the original, the cases we have identified are likely just the tip of the iceberg," an official from the organisation said.
Japan's Justice Ministry has established an expert panel to discuss legal responses to unauthorised use and is expected to compile guidelines as early as this month.
The study examined AI-generated content featuring figures with faces and voices closely resembling those of celebrities and voice actors. Many of the videos were fake, either turning anime into live action or making characters sing using an anime character's voice.
The estimated economic losses were calculated based on the fees associated with using a person's likeness and voice, as well as the advertising value generated by the view counts.
Of the images and videos, 43,138 were suspected of using a celebrity's likeness without permission, while 345 involved voices. By social media platform, TikTok accounted for 43,122 posts, followed by YouTube with 232 and X with 129.
The study's figures are cumulative and include cases that may have been counted twice or more. - Bernama-Kyodo
