# WikiHow and Editorial Perfil Sue OpenAI Over Copyright Training Data

In late August 2026, how-to website WikiHow and Argentinian news publisher Editorial Perfil filed separate lawsuits against OpenAI, alleging the company trained ChatGPT on their copyrighted content without permission, license, or compensation. These cases join over 30 legal actions against OpenAI since late 2023, led by The New York Times, with plaintiffs arguing that ingesting copyrighted material for AI training constitutes mass infringement rather than fair use.

The lawsuits represent a critical escalation in the copyright war because they target two distinct content categories: instructional how-to guides (WikiHow) and international news reporting (Editorial Perfil). WikiHow’s content is precisely the type of task-oriented material that makes ChatGPT useful for step-by-step instructions, while Editorial Perfil’s Spanish-language journalism tests whether US fair use doctrine extends to foreign publishers whose content was scraped without geographic restriction.

What makes these cases particularly dangerous for OpenAI is the scale and specificity of the allegations. Plaintiffs are not claiming incidental copying; they’re alleging systematic ingestion of entire archives—hundreds of thousands of articles—through automated scraping that bypassed paywalls, robots.txt files, and other technical protections. If courts rule this requires licensing, OpenAI’s training data pipeline faces existential disruption, potentially forcing the company to either pay retroactive damages, negotiate ongoing licensing fees, or retrain models on narrowed datasets.

## Fast Facts

- **Filing Date:** Late August 2026 (specific dates vary by jurisdiction)
- **Plaintiffs:** WikiHow (US-based how-to website) and Editorial Perfil (Argentinian news publisher)
- **Total Lawsuits Against OpenAI:** Over 30 since late 2023, including NYT, News Corp, Getty Images, and multiple authors
- **Core Allegation:** OpenAI trained ChatGPT on copyrighted content without permission, license, or compensation
- **Legal Theory:** Copyright infringement (not fair use), with claims of deliberate circumvention of technical protections
- **Potential Damages:** Statutory damages up to $150,000 per work for willful infringement, plus injunctive relief

## Copyright Litigation Landscape

```
+--------------------------------------------------------------------------+
|              OpenAI Copyright Lawsuit Timeline & Plaintiffs              |
+--------------------------------------------------------------------------+
[Late 2023: First Wave]
- New York Times (news articles)
- Authors Guild (book authors)
- Getty Images (stock photography)
               │
      ┌────────┴──────────────────────────────────────────┐
      ▼                                                   ▼
[2024-2025: Second Wave]                            [Late 2026: Third Wave]
- News Corp (WSJ, Times of London)                  - WikiHow (instructional content)
- Reuters (news wire service)                       - Editorial Perfil (international news)
- Multiple individual authors                       - RESULT: 30+ pending cases globally
+--------------------------------------------------------------------------+
```

## Copyright Infringement Claims Matrix

The table below outlines the specific legal theories and remedies sought in WikiHow and Editorial Perfil lawsuits:

| Claim Type | Legal Basis | Alleged Conduct | Potential Remedy |
|---|---|---|---|
| **Direct Copyright Infringement** | 17 U.S.C. § 106 (reproduction right) | Scraping entire archives to train ChatGPT | Statutory damages up to $150,000 per work |
| **Vicarious Infringement** | 17 U.S.C. § 501 (secondary liability) | Profiting from infringing training data | Injunctive relief, profit disgorgement |
| **DMCA Violation** | 17 U.S.C. § 1201 (anti-circumvention) | Bypassing paywalls and technical protections | Additional statutory damages, attorney fees |
| **Unfair Competition** | State law (e.g., California UCL) | Free-riding on plaintiff’s content investment | Restitution, corrective advertising |
| **Declaratory Judgment** | Federal Rules of Civil Procedure 57 | Seeking court ruling that training is not fair use | Precedential ruling affecting all AI companies |

## Real-World Utility & Policy Implementation

### The 4-Step AI Copyright Compliance Playbook

1. **Audit Training Data Provenance:** Conduct a comprehensive forensic analysis of all datasets used to train current and past models, documenting source URLs, licensing terms, and any technical protections that were circumvented. Engage third-party auditors to validate findings and identify high-risk content categories (news, books, code, images).

1. **Implement Opt-Out Mechanisms:** Deploy a standardized machine-readable opt-out protocol (such as the proposed “AI Robots.txt” standard) that allows publishers to signal whether their content can be used for AI training. Honor these signals retroactively by filtering opt-out domains from future training runs.

1. **Negotiate Licensing Frameworks:** Proactively approach major content categories (news publishers, book authors, stock photo agencies) to negotiate blanket licensing agreements that cover past and future training use. Structure deals as revenue-sharing arrangements tied to model usage, creating aligned incentives rather than adversarial dynamics.

1. **Build Attribution Infrastructure:** Develop technical systems that can trace model outputs back to source training data, enabling proper attribution and royalty distribution when specific works are substantially reproduced. This may require embedding watermarks or metadata during the training process itself.

## Actionable Takeaways

1. **Fair Use Doctrine Under Existential Threat:** These cases will test whether US copyright law’s fair use exception—which permits limited copying for transformative purposes like criticism, commentary, or research—extends to commercial AI training. A loss for OpenAI could establish precedent that ingesting copyrighted works for model training requires licensing, fundamentally reshaping how frontier AI is built.

1. **Global Fragmentation Risk:** Unlike the US, the EU’s Copyright Directive explicitly requires licensing for text and data mining (TDM) unless rights holders opt out. If US courts reject fair use while EU law mandates licensing, AI companies face a fragmented global regime where training data legality depends on jurisdiction—a compliance nightmare for models trained on worldwide web crawls.

1. **Settlement Pressure Mounting:** With 30+ lawsuits pending, OpenAI faces accumulating legal costs, discovery burdens, and reputational damage. Even if the company ultimately prevails on fair use grounds, the financial and operational toll may make settlement attractive. Expect confidential licensing deals to emerge before final court rulings, setting de facto industry standards for AI training compensation.

## Everyday Applications & Content Protection Strategies

The copyright lawsuit brought by WikiHow and Argentine publisher Editorial Perfil against OpenAI underscores how instructional, how-to content is transformed by generative models. Here is how everyday content creators, educators, and media managers can navigate this shift.

### How How-To Creators Can Defend Their Audience and Traffic

Step-by-step text guides are vulnerable to zero-click AI summaries. To ensure your digital brand remains indispensable to everyday readers, focus on elements generative search engines cannot replicate:

- **Integrate Original Video & Photographic Evidence:** Document real-world troubleshooting with firsthand photos, short video clips, and original diagrams. AI summaries can extract text instructions, but readers seek authentic human proof that a method works.
- **Offer Interactive Tools & Downloadable Assets:** Supplement written tutorials with downloadable templates, spreadsheet calculators, or code boilerplates that require users to visit your canonical site.
- **Build Direct-to-Consumer Distribution:** Emphasize email newsletters, private community forums, and RSS subscriptions so your connection to readers does not depend entirely on search engine referral algorithms.

### Everyday Copyright Hygiene for Digital Publishers

Small digital publishers and business websites should establish clear legal signals regarding automated AI scraping:

- **Declare Copyright in Machine-Readable Metadata:** Use Schema.org `creativeWork` attributes with explicit `copyrightHolder` and `license` URLs on all published articles.
- **Audit Content Scraping in robots.txt:** Explicitly specify permissions for OpenAI’s crawlers (`GPTBot` for training, `OAI-SearchBot` for real-time search) depending on your commercial objectives.

### Creator Monetization & Syndication Checklist

To future-proof digital publishing revenues against zero-click AI summaries, creators should implement this 4-point operational checklist:

- **Establish Direct Commercial Licensing Inquiries:** Add a dedicated licensing contact link in your site footer for automated AI syndication inquiries.
- **Watermark Original Infographics:** Place high-contrast, branded source attribution on all custom diagrams, charts, and instructional photography.
- **Host Gated Community Discussions:** Encourage readers to contribute user-generated troubleshooting tips and comments that AI bots cannot pre-generate.
- **Audit Referral Analytics Weekly:** Track conversions from search visitors to recurring newsletter subscribers to protect audience lifetime value.