December 22, 2023
TOKYO – Prime Minister Fumio Kishida said at an expert panel meeting on artificial intelligence Thursday that the government plans to establish an “AI safety institute” around January next year, while also expressing intentions to formulate guidelines for AI developers and other businesses by next March.
“International awareness of AI safety has been growing,” Kishida said at the meeting of the AI strategy council, which is chaired by University of Tokyo Prof. Yutaka Matsuo. “We need an organization to conduct research on safety evaluation methods, create standards and carry out other matters.”
The new institute, which will play a central role in ensuring the safety of AI technology, is envisaged to provide support for tasks such as developing software assessing AI risks. The institute will likely be engaged in international coordination as the United States and Britain also plan to set up similar organizations.
The government is also considering introducing a third-party certification system for AI companies, which the new institute is expected to serve as an accreditation body for.
At the meeting, draft guidelines for AI companies were presented. The draft lists 10 principles that they should consider, such as “human-centeredness,” “safety” and “transparency.” The draft also states that human controllability should be ensured, given the gravity of possible infringements of rights that could result from unexpected AI behaviors.
In light of international guiding principles agreed by Group of Seven members based on the “Hiroshima AI Process,” the draft also calls for generative AI developers in particular to introduce technologies for identifying AI creations and clarifying its history of information, among other matters. New technologies, such as Originator Profile, which identifies the original sources of information, are in mind.