• Pascal's Chatbot Q&As
  • Posts
  • In August 2025, two major Japanese media organizations—Nikkei Inc. (owner of the Financial Times) and The Asahi Shimbun—filed a joint lawsuit in the Tokyo District Court against Perplexity AI.

In August 2025, two major Japanese media organizations—Nikkei Inc. (owner of the Financial Times) and The Asahi Shimbun—filed a joint lawsuit in the Tokyo District Court against Perplexity AI.

They accuse the company of: Large-scale copyright infringement - Unlicensed reproduction of paywalled and proprietary content - Harming the credibility and sustainability of professional journalism.


Nikkei and Asahi Sue Perplexity AI for Copyright Infringement

by ChatGPT-4o

🔍 Case Overview

In August 2025, two major Japanese media organizations—Nikkei Inc. (owner of the Financial Times) and The Asahi Shimbun—filed a joint lawsuit in the Tokyo District Court against San Francisco-based Perplexity AI, a prominent generative AI startup. They accuse the company of:

  • Large-scale copyright infringement

  • Unlicensed reproduction of paywalled and proprietary content

  • Harming the credibility and sustainability of professional journalism

The lawsuit demands:

  • An injunction to stop further use of their content

  • The deletion of already-ingested articles

  • Damages totalling ¥4.4 billion (~$30 million USD or £22.2 million)

📄 List of Key Grievances

  1. Unauthorized Use of Content Behind Paywalls
    Perplexity allegedly used articles—including those requiring subscriptions—without obtaining licenses or seeking permission from publishers.

  2. Reproduction Without Compensation
    The plaintiffs argue Perplexity is “free riding” on labor-intensive journalism without contributing to the economic sustainability of the media ecosystem.

  3. Inaccurate Outputs and Misattribution
    Despite citing the newspapers, Perplexity’s AI-generated summaries reportedly misrepresent facts, damaging the publishers’ reputations.

  4. Systematic and Ongoing Violation
    The companies allege that the infringement has occurred “at scale” since at least June 2024.

  5. Erosion of Trust in Journalism
    Asahi and Nikkei highlight the broader threat to democratic institutions, given journalism’s role in disseminating accurate information.

  6. Uncompensated Model Training
    Echoing similar lawsuits (e.g., Dow Jones, News Corp, BBC), the publishers argue their content may have been used in training large language models (LLMs), again without consent or remuneration.

🧾 Quality of the Evidence (as currently known)

Based on public reporting:

  • The complaint itself has not been made public yet, limiting detailed scrutiny.

  • The grievances are consistent with prior litigation against AI firms (e.g., NYT v. OpenAI/Microsoft; Getty v. Stability AI).

  • Claims that Perplexity included false information while attributing sources give weight to the harm argument.

  • No screenshots or technical audits have been published to verify the extent of scraping or LLM training—so these assertions, while credible, are still alleged.

  • The presence of similar lawsuits from reputable entities (Dow Jones, BBC, News Corp) adds credibility to the general pattern of behavior.

Thus, the evidence is strong in its alignment with a documented pattern, but public-facing documentation is still limited, pending court disclosures.

🧩 Perplexity’s Response So Far

  • The company claims it has not yet seen the formal lawsuit, and has requested time for “misunderstandings to be resolved”.

  • CEO Aravind Srinivas has previously defended the company’s practices, suggesting their goal is to “create a better internet” while acknowledging that “publishers need to get paid.”

  • Perplexity has initiated a revenue-sharing program with some media partners (e.g., Time, Der Spiegel, LA Times), and is working on a new monetization model via its Comet browser.

🧠 Lessons for AI Developers & What Should Be Done

 Immediate Recommendations for AI Makers

  1. Obtain Licenses Before Use

    • Do not crawl, summarize, or train on paywalled or proprietary content without an explicit license or opt-in agreement.

  2. Improve Attribution and Transparency

    • If content is cited, make sure it is factually accurate, clearly distinguishablefrom AI-generated summaries, and links back to the source.

  3. Disclose Training Data Sources

    • Implement transparent data documentation (like Model Cards and Data Sheets) to show provenance and whether copyrighted content was involved.

  4. Create Fair Compensation Mechanisms

    • Offer opt-in monetization programs for publishers, including per-query micropayments, licensing fees, or share-of-revenue schemes.

  5. Deploy Guardrails for Quality Control

    • Prevent hallucinations or misrepresentation by cross-checking AI outputs against reliable source data, especially when citing journalism.

  6. Adopt and Adhere to Industry Standards

    • Align with initiatives such as:

      • AI Content Licencing Frameworks (e.g., from WIPO or STM)

      • Do Not Crawl / Do Not Train Directives

      • AI Disclosure and Watermarking Policies

  7. Engage with Regulators and Press Councils

    • Participate in regulatory sandboxes or journalistic AI ethics boards to help co-design governance rules for responsible AI.

🚨 Conclusion: A Warning Shot for Generative AI

This lawsuit is not just about Perplexity—it’s a watershed moment in the battle over who owns and profits from digital content in the age of AI. If generative AI developers continue to rely on scraped, unlicensed, and often paywalled journalism to power their services, they risk:

  • Regulatory backlash

  • Reputation damage

  • Enormous legal liabilities

  • Weakening the very ecosystem of trust and truth that modern democracies depend on

Perplexity and other AI firms must pivot—from scraping to partnering—or face escalating legal and societal consequences.