A breach-of-contract lawsuit is considered more feasible than an IP-based lawsuit, as OpenAI's terms of service prohibit using its content to train competing models. However, enforcing these terms is problematic, as experts argue they may be unenforceable. Additionally, resolving disputes through arbitration rather than lawsuits, and the complexities of international legal enforcement, particularly with China, present further hurdles. OpenAI could have implemented technical measures to prevent such data extraction but at the risk of affecting legitimate users.
Key takeaways:
- OpenAI and the White House have accused DeepSeek of using ChatGPT to train its new chatbot through a process called "distilling," which they liken to intellectual property theft.
- Experts in technology law suggest that OpenAI would face significant challenges in pursuing legal action against DeepSeek, as the outputs of generative AI like ChatGPT may not be copyrightable.
- A breach-of-contract lawsuit is considered more likely than an IP-based lawsuit, but OpenAI's terms of service may be largely unenforceable, complicating potential legal action.
- Even if OpenAI were to win a legal judgment in the US, enforcing it against DeepSeek in China would be complex and uncertain due to differing legal systems and the enforcement of foreign judgments.