Suspected AI-altered image based on Mainichi photo of Kumamoto quake spreads online

A fabricated photograph, seemingly generated by artificial intelligence, based on a Mainichi Shimbun Digital image capturing the aftermath of the Kumamoto earthquake, has been disseminated across digital platforms. The original image depicted a man grappling with the disaster, standing before his destroyed dwelling, his countenance etched with despair.

The fabricated image, however, had been altered to depict the man with a smile and a peace sign emblem on his lips. On July 29, the Mainichi conducted an interview with the man in Kumamoto Prefecture, following the earthquake's aftermath, and published an article in print and online, featuring the photograph of him before his collapsed residence.

The counterfeit version of the photograph subsequently surfaced on Threads, accompanied by disparaging remarks. It was then reposted on X, where it proliferated. Every element of the manipulated image, save the man, mirrored the original photograph with precision, from the collapsed home to the celestial arrangement of clouds. The post also elicited vitriolic comments directed towards the man.

He expressed his anger, stating that it was not merely an affront to him but an insult to Kumamoto, a region where numerous lives were tragically lost. He vowed never to allow such an incident to recur. The Public Relations Unit within The Mainichi Newspapers Co.'s President's Office stated that they had verified the image's high likelihood of being a manipulated version of a photograph published on Mainichi Shimbun Digital.

Such manipulation not only infringes upon their copyright but also tarnishes the dignity of a disaster victim. They deemed this occurrence deeply regrettable and are contemplating potential legal action. False images and videos, produced through generative AI, have proliferated on social media during prior disasters as well. In September 2022, an image alleging damage in Shizuoka Prefecture, inflicted by Typhoon Talas, the 15th typhoon of that year, circulated on Twitter, described as capturing flood damage in Shizuoka Prefecture photographed via drone, with what appeared to be an aerial vista of a submerged town in murky floodwaters.

Yet, the flow of water seemed artificial, and the faces of individuals were obscured. The image was, in fact, a fabrication crafted through generative AI. Analysis by the Japan Fact-check Center revealed AI-generated misinformation also circulated following the earthquake off the eastern coast of Aomori Prefecture in December 2025, encompassing videos illustrating collapsing structures and ground fissures, as well as clips edited to resemble news broadcasts, all bearing watermarks indicating AI generation or labels stating inclusion of AI-generated media.

Additionally, numerous instances have arisen where footage or images from past disasters have been repurposed to falsely illustrate damage that did not occur during an ongoing disaster. The Ministry of Internal Affairs and Communications, along with other entities, caution that various forms of misinformation are particularly prone to spreading during disasters, including misrepresentations of damage extent, unfounded accounts of criminal activity, fabricated pleas for aid, donation scams, and scientifically implausible conspiracy theories.

They urge individuals to scrutinize whether images may have been generated by AI, authenticate social media post sources, and heed expert opinions. (Japanese original by Takayuki Kanamori, Tokyo City News Department) Comment by Tatsuhiro Ueno, professor of intellectual property law at Waseda University. Replicating photographs or content from newspapers or news websites without authorization constitutes copyright infringement.

Furthermore, this case contravenes the right of integrity, safeguarding a copyrighted work against unauthorized alteration, as well as the portrait rights of the individual affected by the disaster. The actions are unlawful on multiple counts, and there appear to be no justifiable circumstances. The post constitutes defamation and could potentially result in criminal liability.