What are the three main principles of the code?

The first requires businesses to disclose the generative AI models they use, training data, and collection methods on their websites with public access (though sensitive information disclosure is optional). The second calls for firms to disclose whether specific webpages are included in training data when requested by copyright holders alleging infringement. The third states that businesses will respond to information requests from system and service users concerned about copyright infringement.

Does this code apply to foreign AI companies?

Yes, foreign businesses providing AI services in Japan are also covered by the "comply or explain" approach.

What happens if a company does not comply with the code?

Under the "comply or explain" approach, companies can choose not to comply if they publicly explain their reasons for not doing so. The code is nonbinding, meaning compliance is not legally mandatory.

AI Business & IndustryHacker NewsAug 20, 2026Aug 21, 2026, 04:03 JST3 min read

Japan to require AI firms disclose training data under new code

Send on LINE

Key takeaway

Japan's government panel has approved a "principle code" requiring generative AI companies to disclose their training data, models, and collection methods to protect intellectual property rights.
The code, based on AI legislation passed in May 2025, uses a voluntary "comply or explain" approach where firms must either follow the principles or publicly justify their refusal.
The measure applies to both domestic and foreign AI businesses operating in Japan and aims to balance copyright protection with technological innovation as concerns grow about AI models being trained on copyrighted material without consent.

3 Key Points

What happened
Japan's government panel approved a "principle code" for generative AI businesses that calls on firms to disclose their AI training data, collection methods, and models to the public. The code uses a "comply or explain" approach — companies can either follow the code or publicly state why they will not, and those complying must announce this on their websites and notify the government. Foreign AI businesses operating in Japan are also covered.
Why it matters
The code addresses growing concerns that texts and images used to train AI models may infringe copyrights without consent. It is built on a law on AI-related technology enacted in May 2025 and aims to balance intellectual property protection with technological innovation. For businesses, the disclosure requirements mean added transparency obligations, though sensitive information disclosure remains optional.
What to watch
The code contains three core principles: disclosure of models and training data methods on company websites; response to copyright holders' requests about whether specific webpages were used in training; and answering information requests from users concerned about copyright infringement. The government will now implement this nonbinding framework across the AI industry.

Ask the AI about this article →

Summaries like this, in your inbox every morning.

Get news like this every morning — free

Context & Analysis

The approval of Japan's "principle code" reflects a broader shift in how governments are addressing the intellectual property risks posed by generative AI. The code emerged from growing concerns about unauthorized use of copyrighted texts and images in AI training, a friction point that has fueled legal disputes globally. By anchoring the framework to legislation passed in May 2025, Japan's government positioned this code as part of a coordinated legal ecosystem rather than an isolated measure.

The "comply or explain" structure is notably flexible — it does not mandate compliance but instead creates a reputational and transparency mechanism. This approach allows companies to weigh the costs of disclosure against public accountability for non-compliance, potentially encouraging adoption without rigid enforcement. The inclusion of foreign businesses underscores Japan's intent to regulate the AI market comprehensively, regardless of where a service provider is based. The three principles form a tiered disclosure system: broad transparency about models and methods, targeted response to specific copyright claims, and reactive support for user concerns — together creating multiple avenues for rights holders and users to challenge potential infringement.

FAQ

What are the three main principles of the code?: The first requires businesses to disclose the generative AI models they use, training data, and collection methods on their websites with public access (though sensitive information disclosure is optional). The second calls for firms to disclose whether specific webpages are included in training data when requested by copyright holders alleging infringement. The third states that businesses will respond to information requests from system and service users concerned about copyright infringement.
Does this code apply to foreign AI companies?: Yes, foreign businesses providing AI services in Japan are also covered by the "comply or explain" approach.
What happens if a company does not comply with the code?: Under the "comply or explain" approach, companies can choose not to comply if they publicly explain their reasons for not doing so. The code is nonbinding, meaning compliance is not legally mandatory.

Hacker NewsRead Original Article

Send on LINE

Also reported by Japan Times Tech

Get the latest AI Business & Industry news every morning

For example, today's edition would include:

DataAgent launches with $10M to auto-fix Kubernetes faultsSiliconANGLE AI · 1h ago
SK Hynix custom HBM boosts inference up to 5.15xDIGITIMES Asia · 1h ago
Taoyuan pitches northern AI data center hubDIGITIMES Asia · 1h ago

AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.

Subscribe free →

Free · takes 30 seconds · unsubscribe anytimeWhat is AIToday? →