The elections to Japan's lower house, held on February 8, 2026, will go down in history not only for their outcome, but above all as the first Japanese elections massively affected by "AI slop" – low-quality content generated by artificial intelligence. As voters increasingly turn to social media for political information, the internet has become a dumping ground for false and dubious reports.
Scale of the problem
On January 30, an anonymous account on the X platform (formerly Twitter) posted an altered election broadcast. The video used text-to-video software to make party co-leaders Yoshihiko Noda and Tetsuo Saito dance. Although authorities removed the post within twenty-four hours, the video received more than 1.6 million views. Major broadcasters reported on the incident, turning a joke into a nationwide debate about fake news.
Professor Harumichi Yuasa pointed out that Japan's Public Offices Election Act does not address manipulated online content. Existing legal remedies therefore remain unclear for candidates harmed by synthetic manipulation. Francesco Cavalli, co-founder of Sensity, added that reliable detection of deepfake videos is becoming nearly impossible.
What AI slop is and why it is dangerous
The Merriam-Webster dictionary named "slop" its 2025 Word of the Year. Greg Barlow described the term as simultaneously fascinating, annoying, and ridiculous. Japanese media have adopted the expression to capture the constant stream of low-quality generative content. Analysts place deepfake videos, chatbot spam posts, and synthetic voice recordings in the same category.
While fake news traditionally referred to deliberate text-based disinformation, AI slop expands this category to include convincing audio and video manipulation. At the same time, some election teams use inexpensive AI animations for legitimate campaigns, blurring the line between creative campaigning and deception for ordinary viewers.
Impact on voter literacy
Recent surveys show that the internet has overtaken television as the main source of daily news in Japan. Approximately 24% of respondents rely on social media for headlines. Heavy dependence on online sources increases the risks when fake news circulates unchecked.
Key figures:
- 46.5% of people read news online daily in 2025
- 1.6 million views the altered video received within 24 hours
- 24% of voters rely on social media for political news
Educational institutions and non-governmental organizations have launched crash courses on verifying synthetic media. Civic technologists are also distributing browser extensions that flag suspicious upload histories. However, compressed election schedules leave minimal room for extensive training.
Algorithms vs. fake news
Platform recommendation systems prioritize engaging content regardless of its veracity. Sensational AI slop therefore often receives preferential algorithmic placement. Algorithmic bias can push altered videos toward undecided demographic groups at critical moments of decision-making. By contrast, dull fact-checks rarely trigger similar amplification.
Professor Shinichi Yamaguchi of the International University of Japan in Niigata Prefecture warned: "The mechanism by which fake news spreads has not changed much since previous elections. But what is definitely different in this election is the advancement of generative AI technology."
The legal framework is lagging behind
Japan's Public Offices Election Act predates social media platforms by decades. Professor Yuasa argues that lawmakers must amend the laws to criminalize the harmful dissemination of synthetic content. However, parliament will not convene before voting closes.
Comparative models include the EU Digital Services Act and Australia's proposed disinformation code. Singapore also enforces removal orders within hours for verified disinformation. Japan instead relies on voluntary guidelines that lack penalties for repeat offenders.
Platform responses and possible solutions
Platform X announced the removal of the viral clip due to misleading content but provided limited transparency metrics. TikTok claims that automated classifiers detect deepfake videos within minutes, a figure that has not been externally verified. YouTube relies on user reports and partnerships with fact-checkers. Experts advocate open databases of removed items to support independent audits. Cryptographic provenance markers could also confirm the authenticity of original broadcasts. Platforms also label official party accounts to support voter literacy efforts.
Yamaguchi urges social media users to maintain a humble attitude when approaching information. He notes that those who have strong confidence in their ability to think critically are actually more susceptible to false information. He also recommends that people check information sources, including whether the sources are original and whether the material was written by anonymous users or experts. Confirming news reports across multiple media outlets also helps verify their reliability.
His team's 2024 survey included 3,700 people across generations. It found that 51.5% of people believed fake news was true. Only 14.5% correctly recognized that the reports were fake. The results showed no difference in vulnerability by age. Yamaguchi warned: "A flood of false content, including AI-generated material, could undermine trust in news and make people more averse to information in general. As a result, there are risks that people will be influenced by simple, forceful messages rather than complex facts. This is a serious challenge for democracy."
Sources: japantimes.co.jp and aicerts.ai
