Deepfakes and Misinformation Cloud Japan’s Recent Election

Misinformation in elections is nothing new, but the level of sophistication has reached unprecedented heights. During Japan’s recent Lower House election, artificially generated content spread rapidly across social media platforms, presenting voters with a new challenge: determining what information they could trust.

An AI-generated street interview video garnered over 400,000 views in the run-up to the February 8 election. The fabricated clip showed a woman criticizing a specific political party. While it originally appeared on YouTube clearly labeled as AI-generated content, subsequent reposts on X (formerly Twitter) removed this crucial context, presenting the fake interview as authentic footage.

This incident exemplifies a growing trend where increasingly convincing deepfakes are being deployed in political contexts. Some AI-generated content even mimicked entire news broadcasts, complete with fabricated reports of political misconduct. While some videos contained small logos indicating their artificial nature, these markers were easily missed by casual viewers.

“Services have emerged that make it extremely easy to create highly realistic content using AI,” explains Professor Taira Kazuhiro of J.F. Oberlin University, who specializes in digital journalism. “The more authentic it looks, the more persuasive it is. And it’s become possible to create and spread videos that damage candidates or political parties.”

The proliferation of such content accelerated after OpenAI released Sora 2 last autumn, making it possible for users to create realistic images and videos within minutes using simple text prompts. While many AI platforms automatically attach watermarks or labels to generated content, these safeguards can often be circumvented by users willing to pay for premium features.

Beyond outright fabrications, misleading election content frequently employs a technique known as “cherry picking” – selectively presenting accurate information without necessary context. This approach leverages factual elements to lend credibility to misleading conclusions.

During Japan’s election, social media posts addressing economic issues illustrated this tactic. One widely circulated claim suggested that abolishing the Children and Families Agency would free up its 7.3 trillion yen budget to either provide 10 million yen to each newborn or eliminate the consumption tax entirely. While the budget figure was accurate, the post omitted that the agency funds essential services like nursery schools, child allowances, and childcare leave benefits, making its elimination impractical.

Similar misleading claims circulated about government subsidies for foreigners. Posts receiving millions of views asserted that Japanese citizens could receive subsidies simply by renting cars to or hiring foreigners. Both claims were refuted by government ministries, which clarified that actual subsidies target businesses developing multilingual services for international visitors and workers – not simply for interacting with foreigners.

According to experts, election misinformation typically stems from three main sources. First, profit-motivated content creators who use sensational claims to drive engagement and monetization. Second, political supporters who exaggerate or distort information to benefit their preferred candidates. Third, foreign actors seeking to influence domestic politics, though no clear evidence suggests this occurred in Japan’s recent election.

As social media increasingly serves as voters’ primary information source, the challenge of separating fact from manipulation will only intensify. Media literacy experts recommend simple verification habits before sharing content: checking who posted the information, examining its source, and reviewing the poster’s history to identify potential patterns of misinformation.

In an era where AI-generated content can be indistinguishable from reality and algorithms prioritize engagement over accuracy, taking time to verify information before amplifying it has become essential to maintaining electoral integrity and informed democratic participation.