The Japanese government is taking steps to secure 'AI sovereignty' to ensure stable use of artificial intelligence (AI) without excessive reliance on specific countries or companies. This move comes after the U.S. government recently changed its policy, which halted Japan's plans for utilizing high-performance AI just ten days after they were announced, revealing the risks of depending on U.S. technology.
According to Asahi Shimbun and Yomiuri Shimbun, the Japanese government approved a revision of its national AI policy, known as the 'AI Basic Plan,' during a cabinet meeting on July 14. This marks the first revision since the initial plan was established in December of the previous year. The rapid advancement of AI technology has raised concerns about cybersecurity and national security, prompting the government to add safety measures and enhance defensive capabilities to the existing focus on promoting use and regulatory adjustments.
The newly emphasized concept in this plan is 'AI sovereignty.' Japan aims to establish a foundation for the stable use of necessary AI without over-reliance on specific countries or companies, while also enhancing its technological strengths in areas where it has an advantage to increase autonomy and negotiation power.
This shift towards AI sovereignty was prompted by recent actions from the U.S. AI company Anthropic, which restricted access to its high-performance AI, Claude Mythos. Mythos is known for its ability to identify security vulnerabilities that humans might overlook, making it valuable for cybersecurity. However, Anthropic limited its general release due to concerns about potential misuse.
On July 2, the Japanese government had anticipated that government agencies and major banks would be able to use Mythos after discussions with U.S. officials. However, just ten days later, the U.S. government reimposed restrictions on foreign access for export control reasons. This decision highlighted the risk that Japanese government and financial institutions might not be able to access the necessary AI technologies. Yutaka Matsuo, a professor at the University of Tokyo, remarked to Asahi Shimbun that this situation symbolizes the increasing importance of AI in international negotiations and national defense.
However, Japan does not aim for complete domestic production of all AI technologies. The ruling Liberal Democratic Party emphasized in a proposal last May that it is neither realistic nor strategic to aim for the domestic production of general-purpose high-performance AI and AI semiconductors, where U.S. companies hold a dominant position. Instead, the focus of policy will be on securing alternative technologies and means in case access is restricted due to decisions made by specific countries or companies.
Japan Bets on Vertical and Physical AI
The Japanese government has identified 'vertical AI,' specialized for specific industries, and 'physical AI,' which operates in the real world, such as robotics and autonomous vehicles, as key areas for competition. By leveraging industrial data accumulated in sectors like manufacturing, healthcare, and construction, along with its advanced robotics technology, Japan believes it can maintain competitiveness without directly competing with U.S. companies in general-purpose AI.
The growth strategy announced by the government last month includes plans to invest 23.1 trillion yen (approximately $211.8 billion) in vertical AI and 10.5 trillion yen in physical AI by 2040. Last year, Japan's private investment in AI amounted to $1.11 billion, ranking 14th globally, which is less than one-200th of the U.S. investment of $285.9 billion.
The government also plans to strengthen its cybersecurity framework. It will utilize high-performance AI to identify vulnerabilities in critical government information systems that are difficult for humans to detect, and it will expand the workforce and capabilities of the 'AI Safety Institute,' which assesses AI safety. Additionally, Japan will enhance cooperation with foreign government agencies and AI development companies to gain early access to the latest AI technologies.
There is also a possibility of introducing penalties for AI operators. The current AI law allows the government to investigate and provide guidance to operators if citizens' rights are infringed due to AI misuse, but there is no basis for penalizing operators who do not cooperate with investigations. The revised basic plan states that to effectively respond to AI risks, related systems, including the AI law, will be 'actively and continuously reviewed.' Yomiuri Shimbun reported that this revision leaves room for strengthening regulations, including the establishment of penalties for non-compliant operators.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.