What About YOUR Data ?
I see a lot of you guys paying for API per-token access for models such as DeepSeek, but would this not also compromise your data? And would your data also land in training for future models, and possibly, almost certainly, also be sold to OpenAI and Anthropic?
If I were OpenAI and Anthropic, with the enormous compute I have, I would definitely put some of it into exposing services for API per-token usage for local models, and then collect the data from people. Or I would just buy it from the API providers directly. This way, I also get to have the data of people who would not share their data with me.
But mostly, I would not leave some data for my competitors that they have and I do not, to prevent exposing any advantage.
Is this not the case? Do those API providers you use have some mechanism to encrypt data and so on?
On the other hand, by the way, if I use the models directly from the original companies, like Qwen, they get my data, and they get to improve their models, and I get better models in the next releases, as opposed to when I use the models through external APIs. i practically punish the maker of the open source models by using their models to create data and give it to others.