The quote from Judge McMahon suggests that while the alleged harm (the use of copyrighted materials without compensation) is currently not actionable under the DMCA...

Question 1 of 2 for ChatGPT-4o: Please see this quote in Peter Schoppert’s blog post "Sweeping statements proven true...":

“Let us be clear about what is really at stake here. The alleged injury for which the plaintiffs truly seek redress is not the exclusion of CMI from defendant’s training sets, but rather the defendant’s use of plaintiff’s articles to develop ChatGPT without compensation to plaintiff,” Judge McMahon writes. “Whether there is another statute or legal theory that does elevate this type of harm remains to be seen. But that question is not before the court today.”⁠

What arguments, statute or theory could the Judge be referring to? List everything you can think of.

ChatGPT-4o’s answer:
The quote from Judge McMahon suggests that while the alleged harm (the use of copyrighted materials without compensation) is currently not actionable under the DMCA, there may be other legal frameworks or theories that could address this issue. Here are some potential arguments, statutes, and legal theories the judge could be referring to:

1. Copyright Infringement

• If a plaintiff could show that the AI model stores and reproduces substantial portions of copyrighted material in its outputs, a traditional copyright infringement claim could be pursued.

2. Misappropriation of Trade Secrets

• If the training data included proprietary or confidential content, the plaintiff might claim misappropriation under trade secret laws.

3. Unjust Enrichment

• Plaintiffs could argue that the defendant was unjustly enriched by using copyrighted works without payment, even if no direct copyright infringement is found.

4. Breach of Contract

• If there was a licensing agreement or terms of use restricting the use of the content for AI training, a breach of contract claim could arise.

5. Lanham Act - False Endorsement or Origin

• If the outputs of the AI are seen as falsely attributed to the original content creators, there might be a claim under the Lanham Act for false designation of origin.

6. Right of Publicity

• In cases involving the use of works that involve personal likeness or identifiable characteristics, claims under right of publicity laws might apply.

7. State-Level Consumer Protection or Deceptive Practices Statutes

• Plaintiffs might argue that using their content without attribution or compensation constitutes unfair or deceptive business practices under state laws.

8. Moral Rights (for Visual Arts under VARA)

• Under the Visual Artists Rights Act (VARA), authors of visual art have moral rights to prevent certain distortions or uses of their work.

9. Breach of Implied License

• A claim could be made that the defendant exceeded the scope of an implied license to use the content for non-commercial purposes only.

10. Preemption by State Law

• State-level statutes that deal with unauthorized data scraping, database rights, or similar protections could apply if federal copyright law does not.

11. Tort of Conversion

• Plaintiffs might argue that their intellectual property was effectively "taken" and used without permission, analogous to the tort of conversion for physical property.

12. Privacy and Data Protection Laws

• If personal data or sensitive information was used in training data, privacy statutes like the California Consumer Privacy Act (CCPA) or GDPR could provide a cause of action.

These theories and statutes highlight various legal avenues outside of the specific DMCA provisions discussed by Judge McMahon. Each theory would depend on the facts of the case and the jurisdiction in which the claim is brought.

Question 2 of 2 for ChatGPT-4o: What other arguments is Schoppert presenting in his blog?

ChatGPT-4o’s answer:
Peter Schoppert’s blog post covers a range of arguments related to the intersection of copyright law, AI, and creative industries. Below are the key points he makes:

1. Fair Use and the Challenges for Copyright Holders in US Courts

• Schoppert highlights the difficulty for copyright holders in successfully arguing that AI companies’ use of their content for training purposes constitutes a fair use violation. He notes that courts seem to require concrete harm, not just a generalized grievance, and cites the case dismissed by Judge McMahon as evidence of this challenge​.

2. The "Brussels Effect" on Copyright Law

• He points to the extraterritorial influence of EU law, particularly in setting norms for copyright holders to reserve their rights or opt-out of AI training. He suggests that this "Brussels effect" could shape copyright norms globally, even where laws differ​.

3. Opt-Out Proposals and Practical Challenges

• Schoppert discusses the UK's proposal for an opt-out system, which drew heavy criticism from the creative industries. He notes the practical challenges of such a system, especially given how widely content is distributed across the internet without the original authors' control​.

4. The Role of Article III Standing in Copyright Cases

• He emphasizes the importance of showing specific, concrete harm to establish standing in US courts. He contrasts this requirement with earlier legal strategies, which focused on unauthorized copying without addressing tangible harm from AI-generated outputs​.

5. Shift in AI Industry Practices

• Schoppert suggests that the AI industry is pivoting from focusing on large-scale data training to improving the quality of data and enhancing performance through methods like reinforcement learning and fine-tuning. He cites Ilya Sutskever's statement about the potential end of traditional pre-training as evidence of this shift​.

6. Copying in AI Training Processes

• Schoppert refutes claims from some legal and technical commentators that no copying occurs during AI training. He describes how data is copied, processed, and tokenized before being fed into models, which can ultimately recreate training data in outputs, undermining arguments that no copying takes place​.

7. Need for Revised Copyright and Data Governance

• He implies that legal frameworks need to evolve to manage emerging issues such as the reproduction of copyrighted works by AI models and the use of synthetic data in AI training. He hints at the potential for more neurosymbolic AI systems and data governance as part of the solution​.

8. Broader Implications for Legal Certainty and Industry Practices

• Schoppert discusses the rising costs and legal uncertainties faced by the AI industry as it scales up its operations. He notes that hyperscalers are investing heavily in infrastructure but warns of less legal certainty regarding copyright and data usage​.

9. The Role of Symbolic and Hybrid AI

• He predicts a shift toward neurosymbolic AI, combining large language models (LLMs) with symbolic reasoning components. This could have implications for copyright, as symbolic components may rely less on verbatim copying of large datasets​.

10. Industry’s Need for Pragmatic Solutions

• Schoppert concludes with an optimistic outlook, suggesting that creative industries and AI companies will eventually converge on pragmatic licensing practices that balance innovation with the rights of content creators​.

Schoppert’s blog reflects his nuanced perspective on how copyright, AI, and technology are evolving and the potential paths forward for resolving conflicts between these domains.