Reality Check: Even if in the past your privacy was safe, eventually your privacy might not be safe, wild     See Less

How does conversions after the training have to do with Training? The conversions are after!?! Right?     See Less

Users should sue to keep their data private. Users also have a stake in the out come different than the NYT.     See Less

WHERE ARE THE LINKS? Thanks.     See Less

But why? Why would news outlets want private ChatGPT chat logs? Why to the New York Times and other news outlets? Can anyone explain why they should have it?     See Less

They shouldn't be saving logs. They absolutely don't have to.     See Less

Where the links you mentioned?     See Less

I really want to come forward and suggest I actually do not want a single company to retain my data and if it is required by legal frameworks especially to prove innocence in a court of law     See More olutely want that to be enforced.

Our relationship with our data has to fundamentally change, it is simply not sustainable for whole populations to be mined for data without a compensatory system in place especially in terms of a core open source framework that any person or ai can inspect the code of to see any malicious intent.

That said working with open source models is very useful although typically I only use Google Gemma and IBM Granite because they have very performant characteristics that are abstractly different.

But it is possible now to run open source models ( there are hundreds of not thousands of them) using Llama.cpp or Ollama ( Ollama is much more simple to work with) on basically Ewaste or machines built up to and around ten years ago and both will work on CPU only. Ollama is now incorporating an experimental Vulkan backend and can utilise RADV and Mesa for older AMD GPU'S down to the lower GCN cards which is really good although this is still not really simple to setup. The open source Vulkan backend means that CUDA or ROCm ( although easier to implement) is not mandatory.

If at some point Nvidia goes bankrupt and CUDA is left unsupported then open source drivers via Vulkan will still be available including for Intel GPU's which are also quite affordable some would suggest.

If you have an old pc or laptop that you do not mind wiping the harddrive or SSD of and you have a spare usb stick to make an Ubuntu Linux boot drive.

Install Ubuntu.

press control+ alt+t. ( Launches the Linux terminal [ don't be intimidated it is a powerful command system but to run llms it is easy to get started with])

Run the update and upgrade commands as follows.

Sudo apt update & sudo apt upgrade -y press enter.

Type the following command
curl -fsSL https://ollama.com/install.sh | sh

Press enter.

Then "Ollama run Gemma3"

The Gemma3 command normally is the quantised 4b model larger or smaller capacity models are in the models library on the Ollama.com website.

If the hardware is limited

"Ollama run Gemma3:1b"

Or search the model list for a model that fits happily in your specific volume of ram which if you do not know can be determined in the Ubuntu System monitor in amongst the main programs.    See Less

I find it hard to believe people do not know that none of this was private to begin with. Where do you think your DATA goes? Nowhere? But then again, I've been surprised before.     See Less

Wow better bring that monthly price down     See Less