Japan Planning New Official Body to Ensure AI Safety; Standards to Be Formulated for Generative AI Creators, Others

Yomiuri Shimbun file photo
The Prime Minister’s Office

The government has decided to establish a new organization as early as January that will be responsible for ensuring the safety of artificial intelligence, it has been learned.

The organization will formulate standards for safety evaluation by such entities as companies developing generative AI. It will also conduct surveys and research on such matters as effective technologies for countermeasures against false information. The U.S. and U.K. governments have already announced the establishment of similar organizations, and Japan, the United States and the United Kingdom will work together toward establishing safe AI.

Prime Minister Fumio Kishida will make an announcement about the establishment of the new organization soon. It will be called the AI Safety Institute, and will be established within the Information-technology Promotion Agency, Japan (IPA) under the jurisdiction of the Economy, Trade and Industry Ministry.

It will bring together a wide range of experts in the fields of AI, cyber security and national security from circles including business, academia and the public sector. Staffers of the institute will also include central government officials.

The new organization will be in charge of creating a mechanism for evaluating the safety of AI. Specifically, the organization will formulate standards for safety evaluation to be used by companies developing AI. It will also help develop software for use in testing by AI companies.

If an independent certification system for AI is introduced in the future, the new organization is expected to become an accreditation body. It will also conduct surveys and research on the safety of AI, including on technologies that can be used to prevent the spread of false information.

Various risks from AI have been argued, including the generation of large amounts of false information, invasion of privacy, and the facilitation of crimes. Development of AI is expected to get into full swing in the future, and how to ensure the safety of AI against various risks has become a global issue.

The Group of Seven industrialized countries have been promoting the “Hiroshima AI Process” to discuss such matters as how AI should be regulated. Earlier this month, the G7 agreed on comprehensive international rules related to AI, rules that call for developers to mitigate risks by assessing safety and conducting tests to confirm safety before launching products on the market.