Proton is not stellar privacy wise either. They claim to be privacy oriented but according to their own transparency report their compliance rate is near 90% and they rarely actively contest the requests. In 2025 data was conceded in over 8000 cases.
–Edit–
In other words they seem to be voluntarily handing out user data to pretty much anyone who happens to ask.
They don’t bother me. The blades are not exposed. I also own a ceiling fan and it doesn’t bother me either, even though it has exposed blades.
Helicopter rotor blades. I’m afraid they will hit my eyes. Even seeing them on a video is very uncomfortable and I have to look away or close my eyes until they’re gone.
Your PC is able to run comparatively small language models with limited number of parameters (a few billions). You can even do that on a modern mobile phone. Some of the more advanced models use hundreds of billions or even trillions of parameters. It’s not just the training. It’s more to do with model complexity.
You can also do weather forecasting on a PC, but the result will not be comparable to a forecast made with a supercomputer.
–Edit–
Typo fixes, because while even the most whimsy LLMs can write, apparently I cannot.
Ok, so all we need is your PC. Maybe you should contact Nvidia and tell that to them.
Same, or in fact any sufficiently private space. The wetness (from water) isn’t a problem, piss (or other bodily fluids, depending on what I was up to and with whom in the bathroom) is the problem. I have unreasonably sensitive skin and get eczemas easily, so I need to keep my basement squeeky clean.
Not really. The infrastructure to run language models is hardware wise demanding, consumes vast amounts of water (cooling) and electricity and requires frequent renewing and added capacity.
Yes it will. None of the current AI companies are profitable, have never been profitable and most likely will never be profitable. Their business model simply isn’t economically sustainable. What is being invested in AI is basically imaginary money in a circle-jerk of companies like Nvidia, Anthropic, OpenAI and others. This artificially bloats their value and there is zero chance in hell that the AI bubble will not burst like nothing ever has before.
This is not to say that LLMs and other fancy jargon generators cannot be useful. They absolutely can. What will probably happen is that people will realize that they are just fancy guessing machines without real understanding behind and learn to use them in that way. I hope at least.
Legislation does not require blind obedience. Requests can and sometimes should be contested. Proton doesn’t seem to be taking a very active approach when it comes to contesting requests.
For example German based tuta seems to be much more active with contests, with reported 75% of user data requests contested/denied, which goes to show that resisting requests can and should be done.
While there are differences between the two companies (tuta being a much smaller player and certainly with its own problems) the difference in compliance percentages is simply staggering.