On Monday, 27 January 2025 the release of an Artificial Intelligence tool named “DeepSeek” raised quite the stir in the US stock market. The tool was branded a major competitor – even superior – to OpenAI’s ChatGPT^1.
OpenAI did not take kindly to this. After all, ChatGPT is a US product and DeepSeek is of Chinese origin. The rivalry between the two ought to be obvious only based on that.
For this to be a story though there is (and there ought to be) more.
Apparently DeepSeek was created with a teeny-weeny budget compared to that used to establish ChatGPT. We’re looking at $5.6 million vs. $100 million – $1 billion^2. A shocking 95% minimum difference. DeepSeek came quite friendly too to consumers, being open-source^3 (anyone is free to copy the code and use it for their own benefit) while ChatGPT is proprietary, meaning that use of its code without authorisation is not allowed.
Now imagine the future of OpenAI if everyone can make their own AI tool working the same as (or better than) ChatGPT? Who would subscribe? Where would business come from? Quite a pessimistic view from an investor perspective, but understandable.
Infringement allegations
OpenAI (and its major funder Microsoft) has come up with some grave allegations. Copyright infringement^4.
Apparently there is a process called distillation, defined as a technique designed to transfer knowledge of a large pre-trained model (the “teacher”) into a smaller model (the “student”), enabling the student model to achieve comparable performance to the teacher model. In effect the knowledge of the large AI tool (in this case ChatGPT) would have been transferred to the smaller AI tool (in this case DeepSeek).
The law of Copyright
While copyright in works of art is generally acknowledged and celebrated, it is worthwhile to know that copyright also subsists in data and software code. Information. Both the US and China’s copyright laws are regulated to a significant extent by a treaty called the Berne Convention. The treaty (which binds South Africa and another 170+ countries) binds a member country to respect other member countries’ copyrights as it does its own.Copyright infringement in the present scenario would constitute DeepSeek exploiting the data and code associated with ChatGPT in any one of the Berne Convention member countries without authorisation being granted by OpenAI.
This means that OpenAI can sue DeepSeek in the US, Europe, China or any other country within which they allege suffering a loss due to the alleged infringement.
Yes, even in South Africa.
As an IP attorney, I can’t help but wish that if a lawsuit is initiated, it happens here in South Africa and is fully litigated here. Think of what it could do for our case law! Not to mention the economic opportunities of a few individuals (as I raise my hand from somewhere in the crowd).
The prospect is quite fanciful but can quickly turn out to be a reality.
But that’s the beauty of Artificial Intelligence, isn’t it?
It makes us dream.
1 https://mashable.com/article/deepseek-ai-vs-openai-chatgpt
2https://www.forbes.com/sites/maryroeloffs/2025/01/27/what-is-deepseek-new-chinese-ai-startup-rivals-openai-and-claims-its-far-cheaper/
3 https://github.com/deepseek-ai/DeepSeek-V3
4https://www.forbes.com/sites/johnwerner/2025/01/30/did-deepseek-copy-off-of-openai-and-what-is-distillation/
Brian Dube, admitted and duly registered Patent Attorney, South Africa. +27766734288 admin@dubeiplaw.co.za.
#Intellectual Property #IPProtection #Patents #Trademarks #Copyright #IPlaw #Designs #CopyrightInfringement #Whatiscopyrightinfringement