200 AI Attack Videos a Day: Japan's PM Sanae Takaichi Caught in a Deepfake Smear Scandal

Last October, Sanae Takaichi defeated Shinjiro Koizumi and Yoshimasa Hayashi in the Liberal Democratic Party leadership race to become Japan's first female prime minister.

Seven months later, the magazine Shukan Bunshun spent three consecutive issues tearing that victory open.

100 to 200 AI videos a day, 70% aimed at one rival

The most jarring part of this AI deepfake scandal is the sheer volume.

Video creator Ken Matsui told Shukan Bunshun that during the September–October leadership race, he was generating 100 to 200 videos a day using AI software — all of it engineered to sway opinion in that intra-party contest.

The targeting was deliberate:

• Roughly 70% attacked Shinjiro Koizumi, casting him as an "incompetent puppet" and a "gaffe-prone disaster"
• About 10% went after Yoshimasa Hayashi, claiming his remarks should "completely disqualify" him
• About 20% boosted Sanae Takaichi herself

Koizumi, her closest in-party rival, absorbed the heaviest barrage.

And the operation didn't stop at the party line. During Japan's lower-house election this February, Matsui — on the same team's instructions — produced videos attacking opposition candidates, branding Constitutional Democratic Party candidate Sumio Mabuchi "an amateur who harms the nation" and claiming Katsuya Okada "lies as easily as he breathes."

67 records, and a confession on camera

Takaichi's team flatly denied everything at first.

Pressed in the Diet, Takaichi herself said: "If you ask me whether I believe a weekly magazine's article or my own secretary, my answer is that I believe my secretary."

But Shukan Bunshun's May 21 report laid out the key evidence: between September 2025 and March 2026, Takaichi's secretary Tsuyoshi Kinoshita and Matsui exchanged 67 relevant messages across texts, encrypted messaging apps, and social platforms.

Then, on May 18, Matsui appeared on the YouTube program NoBorder News and confirmed that he had created and distributed the videos. "I didn't meet Kinoshita in person," he said, "but we held online meetings."

The secretary's contacts are documented; the creator has confessed on camera. From there, the room to deny began to collapse.

The mainstream media pivots, and cracks open inside the LDP

The story spread faster than anyone expected.

Mainstream outlets that had previously given only passing coverage to Takaichi's denials pivoted en masse. The Tokyo Shimbun ran an editorial demanding she explain herself.

The LDP began to wobble, too. Toyo Keizai quoted an anonymous senior party figure: "These claims aren't baseless. If Takaichi denies them, she has to produce hard evidence. Otherwise, neither the opposition nor the public will believe her."

Several opposition parties, including the Constitutional Democratic Party and the Japanese Communist Party, are now preparing a full Diet investigation.

Toyo Keizai cited one political insider's blunt read: the AI smear scandal could cost Takaichi not only the premiership but her seat in the Diet.

The real question AI has put on the table

Until now, the public debate over AI-generated video mostly asked one thing: "Can it fool ordinary people?"

The Takaichi affair poses a sharper one: can an organization weaponize it to wage information warfare at industrial scale?

Consider what 100 to 200 videos a day actually means. A production pipeline that once required copywriters, editors, voice actors, and distributors now needs one person and one computer. Attacking a rival no longer takes a coordinated team — a single prompt can mass-produce hostility.

Japan currently has no clear rules governing the political use of AI-generated content. After this scandal, that legislative reckoning will be hard to avoid.