Japan has launched a large‑scale pilot trial of its in-house generated AI platform Gennai, which is to be accessible to about 180,000 government employees across 39 agencies, Japan's Kyodo News reported. The system is expected to be fully implemented in fiscal 2027. 

While this may appear to be a routine step in Japan's digital transformation, it should raise serious concerns. Given that Gennai's training corpus has been selectively filtered, it risks becoming a new vehicle for carrying and spreading one-sided historical narratives.

Developed by Japan's Digital Agency, Gennai is a generative AI platform built at a cost of 4.4 billion Japanese yen (about 28 million USD). One of its key functions is to help bureaucrats draft responses to parliamentary questions. Kyodo News reported in May that Japanese officials had already used Gennai to prepare replies for a plenary session of the House of Councillors, the upper house of Japan's national legislature.

The values embedded in a technological tool are shaped by its data rules and underlying orientation. Gennai's corpus has been artificially fine-tuned and deliberately skewed to align with the political positions of the current Japanese cabinet. It draws mainly on official government documents, including cabinet resolutions, parliamentary records and official gazettes issued after the enforcement of Japan's current Constitution in May 1947. On the other hand, pre-war archives, documents relating to Japan's surrender, and records of the Tokyo Trials are excluded. Trained on such one-sided materials, Gennai is liable to produce biased outputs. On sensitive issues such as historical perception, territorial disputes and wartime responsibility, it will generate responses that reflect the Japanese cabinet's political stance while distorting established historical facts.

Gennai's assistance in drafting parliamentary responses integrates multiple AI-powered functions, including analyzing questions, retrieving past replies and reference materials, drafting response texts, anticipating follow-up questions from lawmakers, and checking for internal inconsistencies. This means that if materials carrying warped historical perceptions are absorbed into "authoritative" official discourse, AI-generated replies may present biased positions as "objective evidence." The false narratives would then be reinforced and recycled within official institutions, becoming institutionally entrenched and regularly disseminated.

Japan's right-wing forces have played a conspicuous role in advancing Gennai. According to the Digital Agency, the AI system is underpinned by the Act on Promotion of Research, Development, and Utilization of Artificial Intelligence-Related Technologies and the Basic Plan for Artificial Intelligence, both adopted under the cabinet of Japanese Prime Minister Sanae Takaichi.

In the third meeting of the AI Strategic Headquarters last year, Takaichi instructed the government to make full use of "Government AI Gennai." She directed that by May 2026, over 100,000 government employees should be able to use it to carry out their duties more creatively and demonstrate the value of "trustworthy AI" to the public. However, the "trustworthiness" of an AI embedded with skewed historical perceptions is highly questionable.

Moreover, Gennai currently relies on U.S. closed-source models, including Anthropic's Claude. The company explicitly labels China an "adversary country," and its technical alignment is steeped in Western geopolitical prejudices. Gennai is therefore not technologically neutral from the outset; it carries a built-in positional bias.

Separately, Nikkei Asia reported that the Digital Agency has introduced AI systems developed by Japanese firms, arguing that government AI must adopt large language models that "fit Japanese vocabulary and expressions and respect Japanese culture and values." This statement lays bare the intention to embed specific political stances deep within technological infrastructure.

Moreover, the bias embedded in Gennai carries spillover risks. According to The Japan News, the Japanese government plans to promote Gennai in emerging and developing countries in Southeast Asia and other regions. If this biased AI system spreads overseas, distorted historical views could reach wider audiences through technology exports, contaminate global datasets, and mislead the international community's understanding and judgment of the history of World War II.

History cannot be falsified, nor can truth be twisted. From tampering with history textbooks and successive visits by Japanese politicians to the Yasukuni war shrine, which commemorates convicted Japanese war criminals, to now attempting to reshape historical narratives through AI, Japan's historical revisionism continues to evolve in its tactics.

Compared with traditional propaganda, AI-powered dissemination is more covert and efficient, capable of cementing false public perceptions subtly over time. AI should be harnessed for the benefit of all humanity, yet it should not become a new tool for the Japanese government to peddle skewed historical narratives and overturn established historical conclusions.