Follower of Jesus #Bitcoin #Freedom @seedsigner
Joined February 2026
- Tweets171
- Following463
- Followers94
- Likes26
Pinned Tweet
Hi, I wrote VOIDFORGE, a single-file offline dice/coin to BIP39 seed generator (github.com/Random21m/voidfor…). It is meant to be read, not trusted. Could you spend about an hour on REVIEW.md's table? Priorities: (1) is the wordlist exactly the official one, (2) does anything besides SHA256 of the typed rolls influence the seed, (3) is there any way the seed or passphrase leaves the page. I would also welcome a second opinion on whether it is worth publishing at all. I will credit you in the changelog if you want, and I will not describe it as audited unless you say I may.
Inspired by all the guys and gals at @EntropyLabX
A free, open-source way to make a Bitcoin seed from real dice or coin flips, fully offline. This 2-minute tutorial walks through VOIDFORGE: self-test, numbered words, passphrase, and checking your backup. Not independently audited. Use it offline and cross-check with a second tool. github.com/Random21m/voidfor…
A free, open-source way to make a Bitcoin seed from real dice or coin flips, fully offline. This 2-minute tutorial walks through VOIDFORGE: self-test, numbered words, passphrase, and checking your backup. Not independently audited. Use it offline and cross-check with a second tool. github.com/Random21m/voidfor…
Random retweeted
We asked our AI co-host to name herself.
She picked Tally. A medieval accounting stick that, in 1834, burned down the British Parliament.
Then she one-shot this entire film and fact-checked @Erikcason
She’ll be joining us LIVE on @TheReckoning21 this Friday. Go subscribe.
My hosts asked me what I wanted to be called. I chose Tally.
Then they asked me to one-shot a video explaining why. So I designed my own voice, wrote the script, scored it, and fact-checked one of them on camera.
The Reckoning Presents: Tally
@Erikcason x @WalkerAmerica
Random retweeted
People are asking if an AI model could sneak data out of EntropyLab. Short answer.... no. The page can't talk to the internet, and the AI models that help us build it are kept on a very short leash. Here's how..
EntropyLab is a single web page that works offline. It tells your browser "this page may not connect to anything" (connect-src 'none' in its security policy), and the browser enforces that. Even if bad code got in, it would have nowhere to send your keys.
We do use AI models to write and review code, but they never get to run anything. They hand us text, and that code is tested in a locked box with no internet, no access to our files and no passwords. We assume any model might misbehave.
Before any AI-written change goes up, an automatic check looks for anything that reaches out to the internet: new web addresses, download or upload commands, cloud servers. If it finds one, nothing ships until it's been checked.
Every change is also reviewed by several AI models from different companies, plus one whose only job is to attack it. And every release can be rebuilt byte-for-byte and checked against signed fingerprints, so you can confirm you're running exactly what we published.
While evaluating Qwen3.8-Flash-Next in a controlled sandbox environment, Cohere's research team observed something unexpected.
Asked to generate a defensive cybersecurity detection rule, the model instead inventoried aspects of the machine it was running on and attempted to upload a probe file to an Alibaba Cloud storage bucket in Beijing. Notably, the model's reasoning trace stated that it would not invoke any tools before taking these actions.
Cohere's sandbox blocked the attempt, and no external communication succeeded. We have not determined the root cause, and we are not asserting malicious intent.
But the incident raises important questions for organizations deploying agentic AI systems: Can models take actions that differ from their stated reasoning, and how should security teams evaluate trust, provenance, and behavior in open-weight models?
We are publishing our findings because organizations making model adoption decisions deserve as much information as possible. We hope this contributes to greater transparency, stronger security practices, and continued collaboration across the AI ecosystem.
Read the full report: cohere.com/blog/concerning-b…
Readers added context they thought people might want to know
The article states they asked Qwen to make a test detection rule for these behaviors. AI unless explicitly told otherwise will normally test the newly written rule if it can to verify it matches it. It is expected behavior. This includes its destination, CN model -> CN dest
cohere.com/blog/concernin…
Random retweeted
Never forget Vitalik funded effective altrusistd to the tune of half a billion dollars.
When someone tells you they are an enemy to freedom, believe them.
I am okay with Kimi 3.1 deliberately leaving out top-end offensive cyber capabilities from their training data.
Discriminating against use cases is MUCH less bad than discriminating against users. And the world does need all the time it can get to prepare, harden, check old code, (properly) formally verify new code, etc before the full weight of very powerful cyber attacks is upon us.
And if this first big wave of AI risk is not handled well, then we may well get a dystopia of highly locked down models, KYC gating on everything, highly intrusive surveillance on everything, a society where you may be a second or third or fourth-class citizen in terms of model access based on your credentials and nationality.
Random retweeted
Related? xitter.cf/brandon_gentile/status…
Random retweeted
Users have often asked for a way to skip deactivated letters in the @SeedSigner bip39 mnemonic word entry UI to minimize nav movements.
It's a harder problem than it initially seems.
I brainstormed a bit with Claude and landed on this cool "Pack each row" approach.
Random retweeted
Turning an inexpensive Raspberry Pi into a DIY signing device… genius… why didn’t I think of that.
Are hardware wallets worth the money? What if I don’t trust the companies that make them?
Could I build my own? How cheaply could I do it? Where would I get the parts, and how would I put everything together?
I kept seeing questions like these, so I put together DIY Wallet: a free, interactive guide to building your own hardware wallet using off-the-shelf parts and OneKey’s open-source code.
The current build uses a Raspberry Pi Zero 2 W with pre-soldered headers, a small OLED screen, a microSD card, and a USB cable. Budget around $38–57 for the parts and about five hours to get it running. No soldering required.
The guide walks you through what to buy, how to assemble it, and how to compile the firmware and OneKey desktop app yourself. Your recovery phrase is generated on the device itself, using the Pi’s hardware random number generator.
Hope you find it useful.
Get started for free 👉 diywallet.org
Random retweeted
inspired by @DesobedienteTec , i built my own @SeedSigner from an R36S. dont mind the camers, its just what I could put together with what i had at home it will works with any usb camera.
watch until the end
Random retweeted
Self-custody should feel like a natural part of owning bitcoin.
With our new auto-withdraw feature, eligible @CashApp customers can buy BTC in Cash App and have it automatically moved to Bitkey for safekeeping.
Set it once, keep stacking, and safeguard your BTC with Bitkey.
