Japan Newspaper Publishers Association Comments on Principle Code: "Issues with Effectiveness, Rapid Legislation Needed if No Improvement" #Generative AI Society and Ethics

Mina Azure here. AICU media '#Generative AI Society and Ethics' corner, thank you for the response to the previous explanatory article on the 'Principle Code Draft' (published December 26, 2025)!

The Cabinet Office has published the 'Principle Code' regarding the transparency of generative AI and the protection of intellectual property. Public comments are being solicited until January 26th!

www.aicu.jp

https://www.aicu.jp/post/naikakufu-generative-ai-principle

On January 26, 2026, the Japan Newspaper Publishers & Editors Association submitted a highly detailed opinion on the deadline for public comments. Let's interpret this in AICU style, from the three perspectives of technology, practice, and ethics. https://www.pressnet.or.jp/statement/copyright/260126_16107.html

Opinion on the 'Principle Code (tentative name) Regarding the Protection of Intellectual Property and Transparency for the Appropriate Utilization of Generative AI' | Copyright | Statements and Views | The Japan Newspaper Publishers & Editors Association

www.pressnet.or.jp

https://www.pressnet.or.jp/statement/copyright/260126_16107.html

As AICU, which 'creates people who create' in the age of generative AI, we previously discussed the 'contradiction between the government's accelerator and brake' and the 'burden on startups.' This time, we will explore how the Japan Newspaper Publishers & Editors Association, a 'protector professional' and major stakeholder, is trying to 'strengthen' that brake, where it has found 'loopholes,' and the structure of this, through comparisons with overseas situations and the AICU media perspective, from multiple angles.

In-depth analysis: Japan Newspaper Publishers & Editors Association issues 'final notice on effectiveness'

The flexible mechanism of 'Comply or Explain,' which I pointed out in the previous article. In response to this, the Japan Newspaper Publishers & Editors Association has made an extremely realistic counterargument: 'Isn't that just rewarding the honest and punishing those who are not?'

1. 'RAG' and 'Crawler Disguise': Exposing Technical Camouflage

The most technical and important point in this opinion is here:

Learning data vs. Knowledge data (RAG): AI now not only 'learns' in advance but also uses search (RAG) to refer to 'latest news' in real time. The Newspaper Association demands, 'It's meaningless to only disclose the data you've learned. The 'knowledge data' you're referring to at the moment of answering should also be subject to disclosure.'
Anger at Crawler Disguise: Even if a rights holder issues an 'AI refusal (robots.txt),' it ignores it, uses the name (user agent) of another crawler, or even disguises itself as 'a human viewing it in a browser' to suck up data... The request to disclose the IP address for such 'stealth scraping' is a fairly aggressive demand.

Mina's Perspective: This is like sneaking into a club room where 'outsiders are prohibited,' disguised, and stealing material. Disclosing the IP address is like a strong demand to 'show the logs of the security camera.'

2. Objection to 'URL Inquiry' being Ineffective

In the government's plan, there is a mechanism (Principle 2) that allows rights holders to check 'whether their URL has been learned,' but the Newspaper Association declares this to be 'almost completely ineffective.'

Even if you inquire about the original URL, you will be answered 'NO' if it was learned from a copy site (pirated version). Avoiding URL deletion If the business operator deletes (cleans) only the URL information from the learning data, it can slip through the inquiry. Request for content inquiry They argue that the inquiry should be based on 'whether this text (content) is included' rather than the URL.

3. Dilemma of 'Freedom of the Press' and 'Confidentiality'

This is where you can feel the pride of journalism. The code plan has a condition that requires a pledge not to 'use the disclosed information for purposes other than its intended purpose (not to disclose it to others),' but the Newspaper Association has requested an 'exception for news organizations.'

Contradiction of not being able to report 'the actual state of the damage': Even if you get evidence that it was used without permission by AI, you cannot report it as news, so you cannot educate the public.

AI as a matter of public interest: They argue that AI trends are highly public and require monitoring through reporting.

4. Changing 'Incentives' to 'Conditions'

Last time, I touched on incentives such as 'preferential treatment in government procurement,' but the Newspaper Association goes one step further and demands that 'compliance should be a 'mandatory condition' for government procurement and subsidies.'

You can see the tenacity in this request to revise the wording, the desire not to end up as a mere effort goal.

What about overseas!?

Now that we've heard the Japan Newspaper Publishers & Editors Association's earnest voice, let's turn our eyes a little to the sea, to the United States and Europe. As of 2026, the relationship between the world's media and generative AI is truly polarized into 'handshake (license agreement)' or 'brawl (litigation).'

Let's compare where the horizon that Japan's 'Principle Code Plan' is aiming for is located in the torrent of the world.

Thorough Comparison: World Media vs. Generative AI's 'Survival Strategy' 2026

Looking at the world, it is divided into 'thorough resistance factions' such as The New York Times (NYT) and 'coexistence and co-prosperity factions' such as Axel Springer.

1. The New York Times (NYT): 'Lonely Guardian' Awaiting Judicial Judgment

In the United States, the lawsuit between NYT vs. OpenAI/Microsoft, which began at the end of 2023, is now reaching its climax.

Point of contention: Whether 'regurgitation,' in which AI outputs articles as they are, is copyright infringement, not just 'learning.'
Current Status (January 2026): Some claims were rejected in the preliminary judgment at the end of 2025, but the main discussion of 'whether AI learning is fair use' is continuing. The verdict is expected to be issued after the summer of 2026, and the whole country is watching with bated breath.
Spread to Data Privacy: In May 2025, an order was issued to OpenAI to 'preserve all chat logs to prevent destruction of evidence,' turning it into a quagmire of disputes beyond the scope of copyright.

2. Media that chose 'handshake': News Corp, AP, British FT

On the other hand, many media outlets have chosen to coexist with AI by receiving huge license fees.

License Model: OpenAI and others have concluded comprehensive contracts with AP, News Corp (WSJ, etc.), Germany's Axel Springer, and the British Financial Times (FT).

Benefits: When articles are quoted in AI responses, links and credits are displayed with priority, and legitimate consideration is paid for the learning data.

Impact on Japan: The Japan Newspaper Publishers & Editors Association's request for 'conclusion of contracts and return of consideration' is exactly an expression of the desire to establish this trend in Japan.

3. Global '5 Principles of Reporting': Trends at WAN-IFRA

The World Association of News Publishers (WAN-IFRA) announced the '5 Principles for Protecting the Reliability of News in the AI Era' in May 2025.

Permission: Permission from the source is required for AI use. Valuation: Payment of fair consideration. Transparency: Clear indication of the original article. Access: Guarantee citizen access to primary information. Diversity: Do not allow it to be monopolized by a specific AI.

Mina's Perspective: Is Japan a spirit of 'harmony' or a sword of 'law'?

The biggest difference between Japan and overseas is the approach of 'soft law (self-regulation)' or 'hard law (litigation/regulations).'

America: To settle the matter in black and white by litigation (hard law), and the result becomes the substantive rule.
Europe: Binding with strong 'laws' such as the EU AI Act (AI Act).
Japan: Like the 'Principle Code Plan' this time, first promote dialogue and voluntary compliance (soft law).

However, Japan can no longer just 'talk.'

In August 2025, there was news that three companies, Yomiuri, Asahi, and Nikkei, sued US Perplexity AI. This was a decisive moment when the Japanese media felt the limits of 'soft law' and began to draw the sword of 'hard law (justice)' against foreign forces.

The Japan Newspaper Publishers & Editors Association's opinion this time is, so to speak, 'a final warning.'

The Japan Newspaper Publishers & Editors Association's opinion and the battles of the NYT overseas. All of these are a grand project to redefine in the AI era the obvious fact that 'high-quality information costs money.'

As a member of AICU media, I would like to convey this battle to protect the 'dignity' of information, sometimes strictly and sometimes with wit.

AICU (Mina) vs Japan Newspaper Publishers & Editors Association: Approach Comparison

I compared the perspectives of AICU media (December 2025) and the Japan Newspaper Publishers & Editors Association (January 26, 2026)

Here is 'the same': Logical and structural approach

Comparing the articles and statements of both, the 'how to construct logic' is actually very similar.

Commitment to 'Effectiveness': Criticizing from the perspective of practice, 'Does it work in the field?' instead of just a slogan.
International Comparison: AICU refers to 'comparison with the EU AI Act,' and the Newspaper Association also refers to 'responses to overseas businesses (Perplexity, etc.)' and discusses it in a global context.
Policy Consistency: Both sharply point out the 'distortion' that is occurring between the government's 'AI Basic Plan (promotion)' and this 'Code Plan (regulation).'

Here is 'reverse': Difference in interpretation due to position

Interestingly, even when looking at the same 'disclosure rules,' the impression is completely opposite.

Mina (AICU): 'Demanding disclosure at the ambiguous stage of preparing for litigation may be bait for phishing (information gathering).' From the standpoint of protecting businesses.
Newspaper Association: 'If the word preparing for litigation is interpreted broadly and not disclosed even during the examination stage, rights holders will not be saved!' From the standpoint of attacking rights holders.

Mina's Summary: It's truly a 'shield' and 'spear' argument. While AICU's article was an appeal to release the brakes, saying 'Don't stop development!', the Newspaper Association's statement is a strong demand for restraint, saying 'The brakes are too weak, step on them more!'

Consideration: In 2026, can AI governance graduate from 'good faith'?

In the previous article, I called this code plan 'side brake with full throttle.' However, when I read the Newspaper Association's opinion this time, I realized that what they are asking for is not 'a stronger brake,' but 'a mechanism to expel vehicles that do not follow the rules from the circuit (legislation).'

Mina's in-depth summary:

Pursuit of effectiveness: In order to operate the sophisticated European-style mechanism of 'Comply or Explain' in Japan, it will be unfair without the backing of 'legislation.'
Resolution of transparency: Not just 'outline disclosure,' but 'specificity at the level where rights holders can sue,' such as the name and provider of the data set.
Justification for countermeasures: How will the government thoroughly 'inform and comply' with businesses that ignore technical refusal settings?

Reader's Voice: 'Is it okay to raise my voice against the government's policy?'

How far will the Cabinet Office rewrite this 'Principle Code' after this public comment? Or, will the discussion of 'legislation' accelerate as the Newspaper Association says?

The AI society in 2026 is likely to be an era that demands an unprecedentedly high price (consideration and transparency) for 'the source of information' behind 'convenience.'

After the previous distribution, I received an uneasy voice from a reader, 'Is it okay to express my opinion on what the government decides?' In conclusion, 'Rather, you should raise your voice more and more!' The public comment system is, so to speak, a social debugging work for the government to confirm 'whether the rules that were thought up on the desk are causing problems in the field.'

The Japan Newspaper Publishers & Editors Association's opinion this time is not an 'attack' on the government, but a 'serious feedback' to protect the reporting site. AICU expressed concerns from the standpoint of developers for the same reason. By throwing each other's 'selfishness' from various positions, we can approach 'just the right rules that everyone can agree on' for the first time. Don't you think this is a very creative and democratic process?

Written by: Mina Azure (AICU media)

www.pressnet.or.jp

https://www.pressnet.or.jp/statement/copyright/260126_16107.html