A new “comply or explain” framework aims to protect intellectual property without stifling generative AI innovation.
- Transparency Mandate: Japan is adopting a new “principle code” urging both domestic and foreign AI companies to publicly disclose their training data and collection methodologies.
- Flexible Enforcement: Operating on a “comply or explain” basis, the nonbinding framework allows tech firms to voluntarily adhere to the code or publicly justify their non-compliance.
- Three Core Principles: The draft centers on broad data disclosure, direct accountability to rights holders, and responsiveness to end-user concerns regarding potential copyright infringement.
As generative artificial intelligence continues its explosive global growth, a glaring issue has emerged: the unconsented use of texts, art, and data to train complex models. In response to mounting concerns over widespread copyright infringement, a Japanese government expert panel on intellectual property rights broadly approved a pioneering plan this Tuesday.
Rooted in a landmark law on AI-related technology enacted in May 2025, the new “principle code” seeks to harmonize rapid technological innovation with the robust protection of creators’ intellectual property. The framework represents a vital step toward ensuring that the AI models of tomorrow are not built on the uncredited work of today’s artists and writers.
The “Comply or Explain” Approach
Rather than wielding a heavy, restrictive regulatory hammer, Japan is opting for a flexible yet highly transparent mechanism. The government will implement a “comply or explain” approach for generative AI businesses, encompassing both system developers and service providers. Notably, this net also catches foreign businesses providing AI services within Japan.
Under this nonbinding system, companies face a clear choice: comply with the new standards or publicly explain their reasons for opting out. Firms that decide to embrace the code will be required to announce their compliance on their official websites and formally notify the Japanese government, creating a market-driven incentive for transparency.
The Three Pillars of the Draft Code
At the heart of this initiative are three foundational principles designed to peel back the layers of the often-opaque AI industry:
- Public Disclosure: Businesses are required to publish details about the generative AI models they use, their training data, and their data collection methods on their websites. To protect trade secrets and privacy, the disclosure of sensitive information will remain exempt.
- Accountability to Creators: If a copyright holder alleges infringement and makes a formal request, the AI firm must disclose whether specific webpages or materials were included in their training datasets.
- User Responsiveness: The final principle ensures that businesses will actively respond to requests for information from everyday system and service users who harbor concerns about copyright violations.
By bridging the gap between tech developers and content creators, Japan is setting a fascinating precedent in global digital governance.
