The open source label is a problem but not unique to DeepSeek. I think the biggest offenders are the big figures in AI who have conducted a marketing campaign repeatedly mislabeling closed source open weight models as open source. Yann LeCun, Andrew Ng, etc speak open washing lies almost everyday if you follow them. LeCun is especially corrupt since he’s just trying to paint Meta’s Llama positively by pretending it is open source.
But yes this sentence in the DeepSeek license is a huge problem:
> The courts located in the domicile of Hangzhou DeepSeek Artificial Intelligence Fundamental Technology Research Co., Ltd. shall have exclusive jurisdiction of any dispute arising out of this agreement.
Good luck when DeepSeek users try to navigate legal issues, like just yesterday when it was revealed that they’ve been leaking everyone’s conversations to the world.
Definitely not the only model doing this, but this one's a little more complicated. The GitHub repo for R1 just has an MIT license, and most people commenting on this across both tech and general news reports are missing this.
Everyone keeps on reporting that it's covered by the MIT license. I will acknowledge this model is more transparent and visible than proprietary alternatives, but the acceptable uses of the model it is derived from just make it unusable for many.
Even the technical people I'm talking to about this one keep on reporting that it is 100% open. None of these are.
But yes this sentence in the DeepSeek license is a huge problem:
> The courts located in the domicile of Hangzhou DeepSeek Artificial Intelligence Fundamental Technology Research Co., Ltd. shall have exclusive jurisdiction of any dispute arising out of this agreement.
Good luck when DeepSeek users try to navigate legal issues, like just yesterday when it was revealed that they’ve been leaking everyone’s conversations to the world.