this post was submitted on 12 Sep 2026
645 points (98.8% liked)
Technology
88014 readers
2940 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related news or articles.
- Be excellent to each other!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, this includes using AI responses and summaries. To ask if your bot can be added please contact a mod.
- Check for duplicates before posting, duplicates may be removed
- Accounts 7 days and younger will have their posts automatically removed.
Approved Bots
founded 3 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Having dicked around with Local LLMs, they do get better the more they learn about you. But going through what these things figure out is kind of horrifying.
Like looking into the training data they find out, they can infer a lot about you and your methodology and it is creepy. Like really creepy.
I'm not anti-AI by any means. I have a Local LLM Agent in my router to keep an eye on things in case anything fucks up and it's pretty good as my little administrator, but still. It knows a lot about me to the point I had to catch myself to not fucking call it "he" or "they." It's an object but it's an object with a frightening amount of info on me. I just wish the average populace had this level of think of what you are giving up.
If you're not that techy... well first, what the fuck are you doing on Lemmy and on the Tech server? But then I want you to think about this. If someone stalked you for a week, what could they find out about you? How much do you think they would learn?
That is how these huge companies are and especially AI. They're stalkers who manage to learn a lot about you by just observing you.
Can I ask what your setup is?
It's fucking shit. I went that "building your own router" thing but had no idea how to do it. I had an old Mac Mini lying around and so turned that into my router by flashing OPNsense on it and ran into issue after issue.
Here's the biggest issue I had that I know some of you were thinking but I wasn't at the time. OPNsense is built on FreeBSD, which lacks proper support for Apple shit, so the Mac got mad at me every time I tried to flash it. So then I got an idea. Built the router inside a VM in the Mac Mini!
Believe it or not, this actually all went pretty well of getting a vRouter inside of a mac computer and having the Local LLM, I uses Google Gemma tuned to Hermes running Ollama on it, but ran into an issue of the AI agent being painfully, agonizingly slow and revving up my tormented Mac Mini every fucking time I asked it to do anything.
This is because Apple bullshit. I am not qualified enough to answer that, I'm a help desk guy so 50% of the time I talk about IT stuff, I'm confidently bullshitting. Anyways, I got oLMX, gave the local AI agent my API kets to my Mac Mini vRouter, and any time I need anything I can just go into the Mac itself as a computer and talk to it.
I am still trying to find out how to do that thing where you like message the agent through some messaging app as that I have 0 idea how to do. Talking to it through Signal would make my life a lot easier, but sometimes you grow as a person when you reject convenience and make your own life harder.
Hahaha, that sounds like a painful experience, I spent a few years building Apple build systems for mobile games and I hated every minute of it.
Ive heard a few people say "just get a mac mini and run your llm there", and I immediately nope out.
It sounds like you're learning shit from it all, even if you're bullshitting your way through it, some of this random knowledge will help you out later on in life.
I’d argue llms know way more about you than a stalker because you ask them private questions, they analyze you. But the other thing too, this isn’t anything new since people post to social media all the time, it’s just a lot more invasive.
Also, if you couldn’t tell from my reply, I’m not a teche I just found myself here.
LLMs themselves don't, but an external memory system that it reads/writes to might or if someone is reading the chats. But running LLMs locally can be very private.