Category: INDUSTRY NEWS

Driving the news: AI industry groups are urging an appeals court in the United States to block what they call the largest copyright class action ever certified, warning it could “financially ruin” the industry.

• The lawsuit, brought by three authors over Anthropic’s AI training, could swell to seven million claimants, each eligible for up to $150,000 in damages.
  
• Anthropic says that’s hundreds of billions in potential liability, a “death knell” four months before trial.
  

Catch up fast:

• U.S. District Judge William Alsup certified the massive class, relying, Anthropic claims, on “50 years” of experience instead of a “rigorous analysis” of membership and ownership questions.
  
• Anthropic is petitioning the appeals court to reverse the decision, warning that the pressure to settle could wipe out its defenses and set precedent for other AI lawsuits.
  

Why it matters:

• Industry groups say allowing copyright class actions in AI training cases will chill investment, slow innovation, and jeopardise U.S. AI ‘leadership’.
  
• “The technology industry cannot withstand such devastating litigation,” wrote the Consumer Technology Association and Computer & Communications Industry Association.
  

Between the lines:

• Authors’ groups, including the Electronic Frontier Foundation, American Library Association, and Authors Alliance, also back Anthropic’s appeal.
  
• They argue copyright suits are a poor fit for class actions:
  
  • Determining rights for millions of works is messy, orphan works, partial rights, defunct publishers, and deceased authors complicate ownership.
    
  • The court’s opt-out plan risks excluding authors who never hear about the case.
    
  • Google’s past $34.5m effort to build a “Books Rights Registry” shows how costly identifying rights holders can be.
    

The bottom line:
Both AI industry advocates and some author groups warn the current case risks forcing settlements without answering fundamental questions about AI training and copyright, leaving everyone in the dark.

A coalition of American novelists and authors is banding together to safeguard their intellectual property and prevent AI companies from pilfering their works to train language models, thereby preserving the copyright and integrity of their creative output.

“Authors must come together now and negotiate with AI companies as a block,” said novelist Doug Preston in an interview with Fortune.com. 

Preston says he has published at least 40 books, all of which he has discovered have been used by AI firms to train their models without his consent. 

Violation

This violation of his copyright led him to believe that authors and AI developers need some sort of licensing system that protects both parties from copyright infringement. 

That’s where the startup Created by Humans (CbH) comes in. 

Created by Humans was launched in June 2024 and acts as a licensing service for authors. It is a licensing bridge between authors and AI developers. 

“Authors can sign up, claim their works, and note the licenses they’d like to grant to AI developers,” said Jen Singerman, co-founder and COO of Created by Humans.  

Why this matters

Creators of all kinds are increasingly frustrated with how their work is being used by AI firms without their permission. 

According to CbH, over 100 authors have signed up to the service (as of January 2025). 

The Authors Guild, a professional group representing over 15,000 authors, including its former president Preston, has partnered with Created by Humans since October 2024.

“We urgently need to give control back to authors and their publishers, and licensing is the means to accomplish that going forward,” said Mary Rasenberger, Authors Guild’s CEO.

Money for work

Preston notes that AI licensing could be a lucrative opportunity for authors, as AI companies rely on books to train their large language models, and require repeated ingestion of new texts with each model iteration. 

The CEO of Created by Humans, Trip Adler, claims CbH is currently AI licensing a book every five minutes. 

Industry pricing models

The big question is, how much can authors charge AI companies when licensing their work? There are four pricing models currently in  place. 

Flat fee – A one-time payment for access to an author’s catalog (e.g., $50,000 – $500,000+ depending on content volume and uniqueness).

Per-word rate – AI firms sometimes pay per million words (e.g., $1,000 – $10,000 per million words, based on exclusivity and quality).

Revenue share – Authors receive a percentage of the revenue AI firms generate using their content.

Tiered pricing – Authors charge different rates based on usage (e.g., research vs. commercial use).

In November 2024, the Authors Guild disclosed that HarperCollins, a leading publisher under NewsCorp, had entered into an agreement with Microsoft, allowing the tech giant to use select HarperCollins nonfiction titles to train its AI models. The deal priced each title at $5,000 for a three-year period of training rights.

Fundraising 

Created by Humans has raised a total of $11 million in funding, including a $5.5 million seed round led by Giant Ventures. 

Investors include 

• Slack co-founder Cal Henderson
• Emmett Shear, co-founder of Twitch and interim CEO of OpenAI
• Kyle Vogt, co-founder of Cruise Automation
• Dropbox co-founder Drew Houston
• Cameron Yarbrough, Torch’s co-founder. 
• Uncommon Capital
• The pre-seed round, led by Floodgate and Craft Ventures, collected $5.5 million in June, 2024.

Pictured are Created by Humans founders Jen Singerman (left), Trip Adler (centre), and Edward Igushev (right).