LDP Proposal to Add Penalties to AI Law
On April 23, 2026, Japan's ruling Liberal Democratic Party (LDP) formally proposed that the government add penalty provisions to the AI Basic Act. The current AI Basic Act (passed in May 2025) establishes fundamental principles for AI development and use but contains no punitive measures, which has been widely criticized as undermining its effectiveness.
The proposal specifically addresses the growing severity of deepfake pornography and AI-generated copyright infringement, calling for stronger regulatory enforcement.
Current State of Deepfake Regulation
Limitations of Existing Laws
Japan currently has no law that directly prohibits the creation of deepfakes. Victims seeking legal recourse must rely on existing statutes, each with significant limitations:
| Applicable Law | Coverage | Limitation |
|---|---|---|
| Defamation (Penal Code Art. 230) | Public statements damaging reputation | Difficult to apply before dissemination |
| Distribution of Obscene Material (Penal Code Art. 175) | Distribution of obscene images/videos | Creation alone is not punishable |
| Revenge Porn Prevention Act | Non-consensual publication of intimate images | Unclear whether AI-synthesized images qualify |
| Right of Likeness (civil law) | Tort-based damages | No criminal penalty; limited deterrent effect |
Government Response
In January 2026, Chief Cabinet Secretary Hayashi Yoshimasa announced that the government would review measures to address AI-generated sexual deepfakes. The LDP's proposal builds on this initiative, calling for new legislation that would regulate deepfakes from the creation stage.
Free Tool Related to This Article
Statute of Limitations Checker
Try our free simulator related to this topic.
Try for free →AI Training and Copyright Act Article 30-4
Purpose of the Provision
Copyright Act Article 30-4 (amended in 2018) permits the use of copyrighted works without the rights holder's permission for purposes such as information analysis, where the use does not involve enjoying the thoughts or sentiments expressed in the work. This provision serves as the legal basis for using large volumes of copyrighted material as AI training data.
The Line Between AI Output and Copyright Infringement
However, when AI output (generated content) reproduces the expression of existing copyrighted works, Article 30-4 no longer applies, and copyright infringement may be established:
- Training stage: Generally lawful under Article 30-4
- Output stage: If the generated content reproduces the "creative expression" of existing works, it may infringe reproduction rights (Art. 21) or adaptation rights (Art. 27)
- Proving dependence: In practice, proving that an AI user relied on a specific copyrighted work remains a significant challenge
The Agency for Cultural Affairs published its "Perspectives on AI and Copyright" in March 2024, establishing a framework that distinguishes between the training and output stages. The accumulation of case law in this area will be closely watched.
Japan's Approach to AI Regulation
The Japanese government has traditionally pursued its goal of being the "easiest country in the world to develop AI", taking a distinctly different approach from the EU's comprehensive AI Act. However, the rapid proliferation of generative AI has made risk mitigation unavoidable.
Regulatory Outlook
| Area | Current Status | Future Direction |
|---|---|---|
| AI Basic Act | No penalties (passed May 2025) | Penalty provisions under consideration |
| Deepfakes | No direct prohibition | Legislation targeting creation stage |
| Copyright infringement | Training lawful under Art. 30-4; output judged case-by-case | Refined guidelines and case law development |
| Personal data protection | Covered by APPI | Amendments addressing AI-specific risks under review |
Practical Considerations
- Businesses using AI-generated content should establish review processes to check whether outputs reproduce existing copyrighted works
- Deepfake victims can still pursue disclosure of sender information (under the Provider Liability Limitation Act) and takedown requests under current law
- Organizations should monitor developments in AI law amendments and begin preparing compliance frameworks for potential penalty provisions
- Companies deploying AI in their operations must accurately understand the scope of Copyright Act Article 30-4 and implement appropriate controls over training data selection and output management
---
*Houritsu no Mikata Editorial Team | Published April 26, 2026*
