March 11, 2025
TOKYO – Sophisticated fake images and plausible texts created by generative AI are causing confusion in society. There are also many cases in which such images and texts are misused for criminal purposes.
The government has begun to create statutory regulation for AI. However, the details of the envisioned regulation are lacking in effectiveness, and they must be criticized as insufficient.
Serious incidents involving AI are occurring one after another. For a while, social media was flooded with videos that used AI to synthesize the voices and images of famous people promoting fraudulent investments. Many people also reported that they had fallen victim to scams.
When the government solicited opinions from the public as part of its efforts to compile a new Strategic Energy Plan that was adopted last month, a record 40,000 submissions were sent in. Of this number, about 4,000 were believed to have been created and sent en masse by people that oppose nuclear power generation, using AI to create such sentences as “We do not approve of the resumption of operations” at nuclear power plants.
This incident shows that a system designed to reflect diverse opinions in policy has been distorted by the abuse of AI by specific groups.
The bill submitted to the Diet by the government aims to strike a balance between the use and regulation of AI. However, the bill emphasizes that it is based on a policy of supporting the development of AI, with the view that “it is a technology that will serve as a foundation for the development of the economy and society.”
Specifically, the bill calls for the establishment of a strategic headquarters headed by the prime minister to compile support measures for companies that develop AI. The aim is to secure the electric power needed for processing vast amounts of information and to develop data centers equipped with special cooling technology.
At the same time, as a measure to regulate AI, the bill explicitly stipulates that the government will investigate cases in which crimes or other wrongdoings occur — such as making fraudulent ads or creating academic papers through AI — and if there is a problem with the AI product used, the government will issue guidance and advice to the company that developed the AI in question.
There were also calls within the Liberal Democratic Party for penalties to be imposed on unscrupulous AI development companies, but the government was cautious and decided not to do so. Without any enforceable measures, will development companies follow the government’s guidance?
The government was originally cautious about establishing statutory regulation on the grounds that it could hinder AI development. The bill can be seen as a shift in that policy, but its stance of prioritizing development seems to have not changed.
The government and the ruling and opposition parties need to deepen discussion in the Diet over whether this bill will be able to regulate the abuse of AI. They should also consider the option of revising the bill to make it more effective.
In the first place, the main cause of these problems is that the government revised the Copyright Law in 2018 to allow AI to learn from written works, paintings and other copyrighted works without the permission of the copyright holder. Revising that law again is also an important issue to consider.