Former Japanese PM Hatoyama Warns Against 'Poisoning' AI With Historical Revisionism

At the Beijing Xiangshan Forum on September 17, 2026, former Japanese Prime Minister Yukio Hatoyama said he opposes what he described as Takaichi Sanae's "poisoning" of AI and expressed concern that wrong historical views are being built into the Japanese government's AI business support system "Genai." Improper use could make AI a very dangerous tool, he said.

Chinese media investigations have shown that Genai, a government AI system, was piloted by Japan's Digital Agency in 2025 and expanded to about 180,000 government employees in 2026. Japan's Diet allocated 4.4 billion yen in a supplementary budget for fiscal 2025 to support the system. Genai is connected to US AI models such as Anthropic's Claude series, and Japan's Digital Agency has collected around 18 million pages of government data, including cabinet decisions, Diet minutes, white papers, and official gazette data dating back to 1947.

The system has been incorporated into Takaichi's AI strategy, with several right-wing officials involved. Experts quoted in the report warned that AI-generated content from such a system could enter official reports and policy documents and be treated by other models as highly credible sources. They also noted that large models can spread narratives across borders, making contaminated historical accounts more likely to reach international audiences.