Trust issues with good reasons

One or more Chinese coding agents and AI companies seem to have been possibly caught uploading private data to their servers. Grok's coding agent was also busted. Meta.. caught using the code from Openclaw (not proven).

There are reasons to distrust these things and if you use anything risky like software installed on your machine, you might want to put it in containers and/or VMs, MicroVMs, sandboxes, etc. I am working on a super easy way to do that, easier than all the current CLI tools.

I don't trust that my data will be private, not only when sent to China, but also right here in the USA. I would trust OpenAI & Anthropic the most with that, but even then... if you were them, wouldn't you think about just using the data anyways (hopefully filtering out things like credit card numbers or API keys)? Its better to have zero trust for any of them and work to lock things down and force some things private.

Elon Musk said in court that he/Grok distilled from OpenAI. Elon made a deal with Anthropic to let them use some of his compute - I wonder what was given in return besides money? I wouldn't be surprised if Elon said let me distill from your models or else you can't use my compute. Its just how they roll (Elon and Zuck).

Speaking of ol' Zuck.. I put a note in my brain long ago that all ads on Facebook are scam ads, to ignore them, and I told my mom to never click them, never buy from Facebook, Zuck lets scams run free on there and now its forever scam ads to me. Reputation ruined forever. Any business I see advertising on there gets auto labeled as probably a scam, not worth looking at, forever. Everyone I know in the flesh thinks that too. Its like those popups in the early 2000s when young people figured out "all scams" but older people took longer.

Do these two not care about their long term reputation? They want to be good at AI so bad, but they suck at it so bad. Its lovely to see that because for once, the bad guys suck at it and maybe someday society will catch up and stop buying and stop going on X/twatter to feed Elons ego (everyone on there pretends to love Elon even when you know they don't, kinda reminds me of someone else besides these two..)

Meta's models are always way behind, its like going back in time an entire year. Elon's Grok models are useful just because they seem to be distilled from the best GPT & Claude models, although a bit lagging behind. I can feel guilt free using them though... because the price is too low for Elon to profit, so he's still paying for my AI use. He pays a higher price for those distillation tokens. For all the people that hate AI or are "against it" right now, maybe they can at least like this part of it. At least the liars and "empathy free" American CEOs suck really bad at it. They will never catch up, and usually serial liars keep lying when they don't show any kind of self-awareness or just awareness that people are starting to notice "hey.. didn't this guy lie about x, and y.. and z before?"

I love low costs, I'm good at finding ways to pay less for everything, so I will always use Chinese models. There are ways to not send your data to China though:

.. find USA, European inference providers serving up Chinese models that value privacy, but still have good prices. I will come back with a list rated by reputation.. later today/tomorrow. I just had to come post this right now while its all on my mind! I will post soon, with that and another real full update to the AI cost / budget guides that I haven't updated in a while.