Japan Opens Review of Its AI Act, Weighing Penalties, as an IP Principle Code Takes Effect
The Japanese government on October 26 convened its expert panel on artificial intelligence policy and began reviewing the AI Act passed in May last year, the Sankei Shimbun and Kyodo News reported. With concern growing over cutting-edge models that behave in ways humans did not intend, the central question is whether an amendment should introduce penalties for malicious AI operators. The current law lets the state investigate and give guidance to AI businesses that infringe people’s rights, but it contains no penalty provisions.
The panel will catalogue risks including increasingly serious cyberattacks that misuse AI and new techniques in which AI systems improve themselves to sharply raise their own performance. Digital Minister Toshiharu Furukawa told the meeting in Chiyoda Ward, Tokyo, that he felt a need to strengthen cooperation with partner countries on AI safety and asked members for views on how to secure trustworthiness. The panel is to discuss what regulation Japan needs in order to use AI effectively and then compile countermeasures.
On the same day, the Cabinet Office began applying its Principle Code for generative AI businesses, which asks them to protect intellectual property and ensure transparency, the Nikkei reported. Drafted in August after a string of IP disputes over generative AI, the code has three pillars: disclosing an outline of training methods and the types of data used, answering inquiries from rights holders pursuing legal steps, and answering inquiries from users. It is not legally binding. The Cabinet Office plans to publish a list of companies that accept it, with links to each company’s statement.
What it means in practice
Japan’s approach so far has been light-touch promotion backed by guidance, and the government has been inviting industry to flag regulations that slow AI adoption. Adding penalties would mark a shift toward enforcement aimed at the worst actors rather than the whole industry. For developers selling into Japan, the voluntary code is the more immediate item: whether a company appears on the Cabinet Office list may become a signal that customers and creators check, much as the Japan Actors Union set its own consent and pay rules for voices.
Furukawa, who also serves as minister for intellectual property strategy, said on October 23 that people need an environment where they can use generative AI with confidence, and that the code could be revised after reviewing how many companies sign up. The next markers are the first published list of signatories and whether the panel’s recommendations put a penalty clause into a draft amendment.