I'm not saying to trust them, just that they have more of a focus on privacy than the big tech AI clients. Any remote AI that runs on external servers should be questioned unless it's end-to-end encrypted (Proton Lumo is the only one I'm aware of in that regard), but even then, you definitely want to stick to local LLMs if you're able to host them. Otherwise, it's forced to come down to a matter of trust.
Discussion
Not sure if they haters or legit:
https://pivot-to-ai.com/2025/08/02/protons-lumo-ai-chatbot-not-end-to-end-encrypted-not-open-source/
My issue with venice is the "trust us bro" attitude.
Never heard of that site before, and they didn't really cite any sources, so I'd be hesitant to trust that. Not saying it's not true, just that the research I've done so far says otherwise.
As for the Venice thing, you're not wrong, they don't exactly have a way to verify whether not they're telling the truth and that is concerning.
But again, it gets back to the whole idea of not using AI for anything sensitive. I'm all for increasing productivity with AI in different ways but at the end of the day the best thing to do when you need to ask sensitive questions is use a privacy-respecting search engine on Tor Browser or a private browser with a VPN you trust.
No perfect solution, unfortunately, but it is what it is.
I find that a good way to do that is use the models offered by #duckduckgo with #tor At least they do not know who's asking what