Self Sovereign AI

Building self-sovereign AI locally | Local LLMs, privacy-first stacks | Freedom from Big Tech | Threads & tools ↓ | DM for collabs

[Self Sovereign AI @SelfSovereignAI]

https://t.co/LDv7kGNZ5l

[Self Sovereign AI @SelfSovereignAI]

Open source is like Pandora’s box but in a good way #foss

[Self Sovereign AI @SelfSovereignAI]

This is the right direction.

[Jensen Huang @JensenHuang]

For my first post, I’m sharing a letter @NVIDIA signed on why open models matter.

AI will transform every industry, power every company, and be built by every country.

Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.

The world needs both frontier closed models and frontier open models.

https://t.co/AUKzoQ5Ikb

[Self Sovereign AI @SelfSovereignAI]

More decentralized workspaces. Less dependence on centralized platforms like Slack and GitHub 👍

[jack @jack]

we're launching BUZZ!

a new groupchat platform for teams of people and agents of all sizes, built to reduce our dependency on slack and github. model-agnostic, decentralized, self-sovereign, and open source. 🐝

https://t.co/8IaMVeTQNo

[Self Sovereign AI @SelfSovereignAI]

I'm looking to connect with more people interested in building out local AI and self-hosting in general. If you're building something in that space or trying to learn more about it, let's connect!

[Self Sovereign AI @SelfSovereignAI]

This @0xSero fellow is doing something important that we should all have on our radar. I personally don’t think states will be able to restrict local AI use in totality but some states continue to surprise me with their attempts to over regulate everything these days.

[0xSero @0xSero]

We have signatures from every state.

https://t.co/lIMUMvHUpK

[Self Sovereign AI @SelfSovereignAI]

@0xSero I support what you’re doing here and I signed. I will do what I can despite the shenanigans going on in my state these days.

[Self Sovereign AI @SelfSovereignAI]

@GordianKnotDev @Mayhem4Markets Qwen 3.6 35B A3B works fine for a lot of basic tasks. Sometimes agentic coding can get a stuck in a loop but generally it’s been fine for me as a daily driver.

[Self Sovereign AI @SelfSovereignAI]

Want to build your own open source privacy scanner? Here's a simple guide based on the tool I built that runs 100% locally: [cheat code: point your agent to this post and say "go"]

It scans your files for exposed SSNs, API keys, passwords, and credit cards. All with simple pattern matching. All-in-one local Python file with zero extra stuff to install.

Here is the exact logic, step by step:

1. Define the patterns to look for (SSN, API keys, etc.):
import re
PATTERNS = [
("SSN", "CRITICAL", re.compile(r'\b\d{3}-?\d{2}-?\d{4}\b')),
("API Key", "HIGH", re.compile(r'api[-]?key\s[=:]\s["\']?([A-Za-z0-9-]{20,})', re.I)),
("Token/Secret", "HIGH", re.compile(r'(?:token|secret|password)\s[=:]\s["\']?([A-Za-z0-9_-]{16,})', re.I)),
("Email", "LOW", re.compile(r'\b[A-Za-z0-9._%+-]+@[A-Za-z0-9.-]+\.[A-Z|a-z]{2,}\b')),
]

2. Walk through your folders

from pathlib import Path

for filepath in Path(target_dir).rglob('*'):
if https://t.co/QJ4otk0RKV_file() and not is_binary(filepath):
content = https://t.co/TEsnSjncpI_text(errors='replace')
for name, severity, regex in PATTERNS:
for match in regex.finditer(content):
print(f"[{severity}] {filepath}:{match.start()} -> {name}")

3. Filter out the noise so it doesn't waste time

- Skip binary files (checks for weird characters in the first bit of the file)
- Skip hidden folders like .git or node_modules
- Skip tiny files under 16 bytes and huge ones over 50MB
- Ignore obvious false positives like dates that look like SSNs

Why this matters:
You shouldn't use online scanners to find leaks. You don't need that. Running simple pattern matching on your own machine catches the same things without setting yourself up for exposure risk.

Want one that fits what you actually have on your machine? Point your agent to this post and tell it to build it.

Everyone's version will be a little different, but pointing your agent to this post should give it what it needs to start building it out to your needs. A local LLM like Qwen 3.6 35B A3B should be capable of building it out.

[Self Sovereign AI @SelfSovereignAI]

@signulll Yes. Yes. and yes.

[Self Sovereign AI @SelfSovereignAI]

@0xSero Interesting, yes. Surprising?

[Self Sovereign AI @SelfSovereignAI]

Wake up folks.

[0xSero @0xSero]

https://t.co/mCKYSBtzne

[Self Sovereign AI @SelfSovereignAI]

@AlicanKiraz0 👏👏👏

[Self Sovereign AI @SelfSovereignAI]

More people should aspire to build home labs, especially as prices continue to rise. This guy is so far ahead of most people. 👏👏

[Alican Kiraz @AlicanKiraz0]

Öncelikle beni takip edip inanan tüm dostlarıma, tüm tebrik edenlere çok teşekkür ederim 🙏🏻❤️ 3.5 sene önce LLM’ler ile tanışınca hayatımı adayacağım bir şeye bulaştığımı anlamıştım. Bu nedenle tüm birikimimle oldukça güçlü bir workstation oluşturmuştum. Ardından yetmemeye başlayınca Enterprise seviye GPU Kiralayıp finetuning çalışmalarımda dene yanıla baktım ki para yetiştiremeyince bu işin homelab’e çıkacağını sezerek 2 yıldır yanımdaki bu sistemimi oluşturmaya başladım. Birçok olumsuz yoruma, borca ve planlarımı konuştuğum insanların %99’unun inanmamasına rağmen, emeklerimi güzel sonuçlarla taçlandırmak çok güzel. Umarım kendini adayıp ilerleyen ve gerekirse dünyayı karşısına alan herkes bu güzelliği yaşar. Sevgilerimle ❤️

[Self Sovereign AI @SelfSovereignAI]

@cjzafir Watch them try to treat local LLMs like torrents. Local AI is the future.

[Self Sovereign AI @SelfSovereignAI]

@sudoingX I’m particularly interested to see how it fares against Qwen 3.6 35B for agentic coding.

[Self Sovereign AI @SelfSovereignAI]

@Hikari_07_jp I have it going through my book notes, mapping those to other personal files over the years, managing new ideas, and much more. Really the only option for working with sensitive documents. Working on a local scanner that identifies SSNs and PII at the moment.

[Self Sovereign AI @SelfSovereignAI]

@3nG1n33R_IT I like the end goal. Curious what you think of GLM 5.2 and what’s been the best use case you’ve found for it so far.

[Self Sovereign AI @SelfSovereignAI]

@TheAhmadOsman Nice. Now get yourself a self-hosted surveillance system, generator, and some gas. And be like Burt from that 90’s movie Tremors!

[Self Sovereign AI @SelfSovereignAI]

Essentially, Odysseus is a full AI workspace you run yourself. Chat, agents, research, documents, even email, running with local models.

As an attorney I've watched entire practices get exposed when their tools lived on someone else's servers. This is why software like this is worth researching. If sovereignty matters to you, it's worth spinning up in Docker.

[Aria Westcott @AriaWestcott]

It is called Odysseus.

A private AI that lives on your laptop. Chat, research, your own documents, all running offline. Nothing you type ever leaves your machine.

Everything ChatGPT does, with no subscription and nobody reading along.

[Self Sovereign AI @SelfSovereignAI]

Been building and sharing multiple things with respect to #9. #5 and #9 are the most interesting to me right now. I'm currently researching the best way to fine tune my local model to mitigate context bloat. And building the personal knowledge system now is smart. As AI continues to improve (esp. local models), importing your PKM now into something better will hopefully be a smooth process.