I don’t know if Deepseek did it or did not.
But one of the things we will see over the next few years, starting today, is the eventual closing up of anything “open source” when it comes to AI, LLMs, training data, anything and everything related to building better models.
Not what is already in the public domain, but definitely an eventual drying up of any “open source” releases. But I wonder how distillation of model knowledge can be prevented.
I installed Linux for the first time in 1996 (thanks to PC Quest, March ’96) and I really, fervently hope this closing of tech doesn’t catch. The benefits of Open Source are astronomical.
But AI is going to become the basis for future sovereign defensibility. It will be closed off. Urgently.
Either way, the deepseek training damage has been done.
