TL;DR: AI companies scraped the entire internet, including your content, to train their models. They're now worth hundreds of billions. Courts are deciding whether this is "fair use" or theft. The New York Times is suing OpenAI. Getty Images is suing Stability AI. Anthropic settled with authors for $1.5 billion. Meanwhile, Disney invested in OpenAI and licensed its characters. 2026 will see landmark rulings that determine whether AI companies must compensate creators, or whether everything you post online is free training data.
What changed (October 11, 2026): Judge Araceli Martínez-Olguín of the U.S. District Court for the Northern District of California granted final approval of the Bartz v. Anthropic class action settlement on July 20, 2026, locking in the $1.5 billion agreement at roughly $3,000 per work (four times the $750 statutory minimum for ordinary infringement). On November 4, 2025, Mrs Justice Joanna Smith of the UK High Court issued the first major foreign ruling on the dispute: in Getty Images v. Stability AI, Getty largely lost its copyright claims after abandoning primary infringement allegations during trial, with the court dismissing its remaining secondary copyright claim and characterizing Getty's narrow trademark win as "extremely limited in scope." Authors Guild v. OpenAI (S.D.N.Y. 1:23-cv-08292) and New York Times v. OpenAI remain pending; a March 9, 2026 decision in the Authors Guild matter is on the docket and the New York Times case continues toward trial.
What Happened
AI companies built their fortunes on your content:[1]
- Web scraping: Bots crawled billions of web pages, copying text, images, and code
- Books3 dataset: 196,000+ pirated books used to train language models
- LAION-5B: 5.8 billion image-text pairs scraped from the web
- GitHub Copilot: Trained on public code repositories, including copyleft licensed code
- ChatGPT: Trained on Common Crawl data covering much of the web
Nobody asked permission. Nobody paid. The companies argue this is "transformative fair use."
Major Lawsuits
NYT vs OpenAI
The New York Times alleges ChatGPT can reproduce its articles verbatim. OpenAI claims fair use. Trial expected in 2026.
Getty vs Stability AI
Getty Images claims Stable Diffusion copied millions of images. Stability argues transformative use.
Authors Guild vs OpenAI
Authors including John Grisham, George R.R. Martin, and Jodi Picoult allege "mass copyright infringement."
Music Publishers vs AI
Warner Music Group settled with AI music startups Suno and Udio. Licensed models coming in 2026.
The Licensing Pivot
Some companies are now paying, after getting sued:[2]
- Anthropic: $1.5 billion settlement with authors
- OpenAI + Reddit: Licensing deal (undisclosed value)
- OpenAI + Associated Press: Licensing deal
- Disney + OpenAI: Disney invested and licensed characters for Sora video generator
- Warner Music + Suno/Udio: Settlement and licensing agreement
The pattern: companies with resources can negotiate licenses. Individuals, small creators, and the public domain get nothing.
The Fair Use Debate
Fair use is the legal doctrine allowing limited use of copyrighted material without permission. Courts consider:[3]
- Purpose and character: Is the use transformative? Commercial or educational?
- Nature of the work: Is the original creative or factual?
- Amount used: How much of the original was copied?
- Market effect: Does the use harm the original's market?
AI companies argue: Training is transformative. The models don't reproduce originals: they create new content. This is like how humans learn.
Rights holders argue: The entire works were copied. The models compete with originals. This is industrial-scale infringement.
Some judges have called AI training "quintessentially transformative." Others worry it could "undermine creative industries." The cases will likely reach the Supreme Court.
The Privacy Angle
Beyond copyright, there's a privacy problem:
- Personal data in training: AI models contain information about real people scraped from the web
- No deletion possible: Once trained into a model, personal information can't be removed
- Memorization: Models can sometimes reproduce exact training data, including personal details
- Gmail lawsuit: A class action alleges Google used private Gmail content to train AI without consent
GDPR gives Europeans a right to deletion, but that right may be meaningless for data baked into AI models.
Is Scraping Legal?
Web scraping occupies a gray zone:[4]
- Public data: Generally legal to scrape publicly accessible content
- Terms of service: Violating website TOS may create liability
- Technical circumvention: Bypassing access controls may violate computer fraud laws
- robots.txt: Ignoring robots.txt is bad faith but not clearly illegal
Lawsuits are testing these boundaries. Reddit sued Perplexity AI for allegedly violating scraping policies. Google sued SerpApi. Outcomes will shape what's permissible.
Emerging Regulations
States are beginning to regulate AI:[5]
- Texas TRAIGA (Jan 2026): Bans certain AI uses, requires disclosures for AI in government and healthcare
- Colorado AI Act (June 2026): Prevents algorithmic discrimination by "high-risk" AI systems
- EU AI Act: Requires transparency about training data for high-risk AI systems
No comprehensive federal AI law exists in the US. Copyright law wasn't designed for this. Courts are improvising.
What Content Creators Can Do
Check robots.txt
Add directives blocking AI crawlers (GPTBot, CCBot, etc.). It's not legally binding but creates evidence of intent.
Watermark Images
Watermarks make scraped images less usable. Not foolproof but raises friction.
Monitor for Copying
Tools exist to detect if your content appears in AI outputs. Document instances for potential claims.
Join Collective Action
Authors Guild, music guilds, and other organizations are filing collective suits. Strength in numbers.
Consider Licensing
Some platforms enable opt-in licensing (Shutterstock, Adobe). Compensation is minimal but exists.
Document Everything
Keep records of when you published what. You may need to prove you created something before a model was trained.
Platform Policies
Where your content lives matters:
- X (Twitter): Policy allows training AI on posts (including yours)
- Reddit: Licensing deal with OpenAI covers user content
- Meta: Using public Instagram and Facebook content for AI training
- DeviantArt: Opt-out available after backlash
- Tumblr: WordPress explored AI licensing deals with user content
Read terms of service, but understand that platforms can change them. You agreed to terms that probably didn't exist when you signed up.
The Bottom Line
If you've posted anything online (a blog, photos, code, comments, reviews) AI companies probably copied it. They used your work to build products worth billions. They didn't ask. They didn't pay. They argue it's legal.
Courts will decide in 2026 and beyond. The New York Times case, Getty case, and author lawsuits could reshape the AI industry. If fair use claims fail, companies may need to license or remove training data. If fair use claims succeed, everything online is fodder.
Meanwhile, the biggest creators are getting licensing deals. Disney gets paid. Warner Music gets paid. Individual creators get nothing, except their work used to train systems that may replace them.
This is the largest intellectual property dispute in history. The outcome affects every person who creates anything on the internet.
Update (October 11, 2026)
The biggest AI training-data verdict since this article ran is now a settlement, not a ruling. On July 20, 2026, Judge Araceli Martínez-Olguín of the U.S. District Court for the Northern District of California granted final approval and entered final judgment in Bartz v. Anthropic, the class action brought by authors Andrea Bartz, Kirk Wallace Johnson, and others. The $1.5 billion settlement works out to roughly $3,000 per work, four times the $750 statutory minimum for ordinary copyright infringement. The settlement releases only claims tied to Anthropic's acquisition and copying of works through August 25, 2025; it does not resolve pending claims about the use of pirated Books3 datasets or future training.
The first major foreign judgment in the space is also a setback for rights-holders. On November 4, 2025, Mrs Justice Joanna Smith of the UK High Court issued judgment in Getty Images v. Stability AI. Getty abandoned its primary copyright infringement claims during trial, and the court dismissed what remained of the secondary copyright claim; Getty won only a narrow trademark infringement finding, which the judge herself characterized as "extremely limited in scope." For the U.S. cases, Authors Guild v. OpenAI (S.D.N.Y. 1:23-cv-08292) and New York Times v. OpenAI remain pending; a March 9, 2026 decision appears on the Authors Guild docket, and the New York Times case continues toward trial with the fair-use question unresolved.
The combined picture: the only fully resolved U.S. matter against a frontier AI company settled without a fair-use ruling, the only fully resolved foreign matter went against the rights-holder on the merits, and the two cases that could produce an actual fair-use precedent are still pending. Individual creators and small publishers have not yet gotten a court ruling on whether their work is being taken without compensation.
Sources: Authors Guild. "Court Grants Final Approval of $1.5 Billion Anthropic Copyright Settlement." July 20, 2026; Copyright Alliance. "What to Know About the $1.5 Billion Bartz v. Anthropic Settlement"; Cleary Gottlieb. "UK High Court Issues Landmark Ruling in Getty Images v. Stability AI." November 4, 2025; UK Courts and Tribunals. "Getty Images v Stability AI." Judgment PDF, November 4, 2025.