Quote from: Parabellum118 on June 23, 2025, 06:25:47 AM

Hello everyone, I think I have a problem!

I read an article saying that with enough transactions and a nonce bias of just 1, it's already possible to recover a private Bitcoin key 😯
ECDSA Cracking Methods (April 2025)
A survey paper by Edinburgh Napier University (William Buchanan et al.) summarizes that nonce biases, weak nonce selection, or even a single broken bit can be enough to recover a key using lattice algorithms like LLL.

So I thought, let me take my transactions, and with the help of ChatGPT I created a script based on that study. Before that, I extracted the RSZ data from the transactions using the tools from iceland2k14/rsz and double-checked them with 2coins.org/RSZ-Signature-From-Tx, and the data is correct.

I then converted the RSZ data to binary format, and during the first check, I already had a bit bias of 8 across 12 transactions 🤨 — in the upper area (MSG). I dug deeper and had ChatGPT generate a script that runs the analysis based on the study. As a result, I found that my transactions actually show a very strong bias, and R correlates with S and Z — theoretically, someone could recover my private key.

Then I tried it with a custom tool created by ChatGPT. Luckily, I wasn't able to recover the private key — I did get a lot of addresses, but mine wasn't among them. Still, I'm feeling uncertain.

There are a lot of RSZ recovery tools on GitHub, but most of them require a fully known nonce. Besides the fact that I've already moved my BTC to a new wallet — which tools are actually capable of recovering a private key with a bit bias of 4 to 8?

No quantum computer needed — it seems to work surprisingly fast. The study doesn't offer tools, it's just a paper.
Since I know the private key for my address, I’d like to try it myself.

Anyone here have experience with this?

Can you share the article paper link?

Quote from: Parabellum118 on June 23, 2025, 06:25:47 AM

I then converted the RSZ data to binary format, and during the first check, I already had a bit bias of 8 across 12 transactions 🤨 — in the upper area (MSG). I dug deeper and had ChatGPT generate a script that runs the analysis based on the study. As a result, I found that my transactions actually show a very strong bias, and R correlates with S and Z — theoretically, someone could recover my private key.

I am all ready to try converting the RSZ data to a binary format. No bias leak. Can you share your code because I went to test how your bias code was working?