Well, E2EE was just a guess on my part.
If computer manufactures start putting a physical chip on the mobo to start spying on us, I will never buy a new computer again. This AI garbage is getting way out of hand. According to the constitution, I have the legal right to privacy. If that is to mean nothing now, then I will stick with the computer I got during the pandemic, and never buy another again.
You don't see me running Windows 11 24H2 for the same reason. Vote with your wallets people! If they are gonna start doing this garbage, let companies know, we have no problem with them going out of business due to too low of profit.
Companies need to make products for us if we are to buy them, not for their own bottom line. No spying AI chip in my 10th generation machine!
I have only good experience with Protonmail 
We can always buy linux machines, system76, framework, tuxedo and so on. I highly doubt those will get ai chips.
Well, I have some good news for you. These chips are almost always (and entirely) dependent upon the software (OS). My new "copilot PC" laptop came with an NPU, and here's what I found:
- The dreaded copilot button launches copilot in a wrapper, which makes sense, since copilot is cloud-based. The NPU sits at 0% use.
- I tried using local LLM (both on Windows using LM Studio and on Linux using Alpaca), and none used NPU.
- The Linux system didn't even report the presence of the NPU. I later found out that only some models can take advantage of these NPUs. So, with time, more local LLMs will be able to take advantage of the chip.
- The only use right now, at least on Windows, seems to be in background blurring when the webcam is on and similar "AI-based on-device tweaking."
The short version is that the chip isn't likely going to spy on you. It will be an OS-level decision. However, the AI hype is too big, and I've seen discussions about AI integration into some Linux distros (I've been lurking in Fedora forums and I think they (or Red Hat?) are considering some on-device AI integrating into the OS).
This is beyond scary - it's truly disturbing! I had wondered if there was some kind of language issue. Say, for example, one user has US English installed and uses a US keyboard, but the other uses a Spanish keyboard but set to British English (that's one option I frequently use), could there be a mis-translation? However, I can see that can't be the case, especially if there is a whole sentence missing. I shudder to think what could happen under certain, sensitive circumstances.
I do use Proton mail but as spam accounts.
Actually... In the time since (Both using US English), we exchanged email passwords and logged into each others accounts. Then looked and compared.
We found a great many emails that were missing sentences, words and even whole paragraphs.
Neither of us had previously noticed, simply because... how would you know something is missing?
I then contacted another person, both of us using Protonmail and we examined ours.
We both found several instances of words containing extra letters in the Sent items, blank emails that originally had content, emails that contained some content, but not all.
Since many of these were innocuous messages like "here is my thought on that issue" or "Today, I have this scheduled" - the recipients had no reason to suspect anything was missing until we found it later.
And seeing a typo is not unusual.
In one case, the recipient thought I sent the email early without finishing a thought - I had no idea that their copy did not match mine and he had not felt it was important to ask about the unfinished sentence.
So apparently... this has been going on - a lot - with multiple users. The only correlation I can find is it only seems to happen when both parties are using Protonmail.
Gmail, Yahoo etc recipients receive the full unaltered message.
I cannot post screenshots of these due to confidentiality of others messages. But I did save a large number of screenshots. I am absolutely blown away. I have never seen anything like this.
Wow!
I've never even heard of anything like this before - it's beyond weird.
Thanks for sharing, @Aravisian . I'll certainly bring this up if anyone I speak to mentions Proton. It;s also given me pause for thought as I was going to use Proton as my main password manager. Your and @zenzen 's experiences have made me think twice... I'll be thinking about what alternatives there are that are practical for me to implement.
I have defended Protonmail and endorsed it on this very forum. I have forgiven a lot of their "oopses" over time.
Seeing what I saw earlier today: Absolutely 100% Not.
That the messages are garbled or missing content is something that they should never have passed onto the public. Proton deserves no consideration, nor benefit of the doubt.
What you describe is 100% unacceptable. Do you suppose that messages are being turned into a string of tokens representing your content, then inaccurately re-assembled? I can't think of how any normal encryption method would create a lossy message.
I have considered Proton a few times recently for their combined service. Thank you for the warning.
Unfortunately, I don't think this will be true for long. Framework has already launched a desktop optimized for "local AI inference". It's not a dedicated AI chip, but one made by AMD with that exact purpose, so it's not too far fetched that this will become a reality if it has success.
This only makes things even weirder... but you're right, we could try to come up with potential causes to this issue, but it shouldn't ever happen. It's the very core of their business!
Years ago I had a protonmail acc and used it's mail client. I found it to be unreliable in fetching messages and some other things I wasn't happy about. So ditched that years ago.
Nowadays I just use thunderbird , I'm not crazy over it ,but it works the best out of MANY clients I've tried.
This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.