The Japanese government confirmed on the 11th that it is considering requiring AI-related industries to assume safety assurance obligations and establishing new penalties.

This plan is the direction set by a non-public working group to discuss the AI ​​legal system. A complete set of system adjustments, including legal revisions, will be announced before the end of the year.

The threats addressed are very specific: AI being misused for cyberattacks, and AI running out of control. Japan’s current AI-related laws focus on “promotion” and are basically non-enforceable. This increase in penalties is mainly to deal with malicious businesses that refuse to cooperate with negotiations.

The most critical question still remains: how to define the scope of application, and whether large American IT companies are included, which is still being sorted out.

The significance of the penalty is to change "please cooperate" into "must cooperate". In the past, when the Japanese government encountered companies that were unwilling to negotiate, it basically had no cards to play. This time, it wants to prepare the tools first.

This week, the pace of global regulation is almost at the same point: California sued OpenAI for losing control of the robot, South Korea's ruling party plans to ban AI cheating from taking exams for 10 years, and Japan is preparing to impose obligations on AI companies. If the plan at the end of the year really includes major overseas manufacturers, Japan will become another major market after the European Union that directly imposes penalties on American AI companies.