What makes this even harder for me is that I tried to speak about what happened before, and my previous topic was deleted. Whether there was a legitimate reason or not, it left me feeling like I had no voice.

Maybe it's just a coincidence. Maybe I'm reading too much into it. But when you've just lost years of savings, it's difficult not to question everything.

This forum is one of the largest Bitcoin communities in the world. If the person responsible for taking my coins is active anywhere, there's a chance they've visited this forum or will someday. That's why I wanted my message to exist here.

I'm not accusing anyone on this forum. I have no evidence to do that. I just wanted there to be a chance that the person who now controls those coins might eventually read the words of the family they affected.
Lost 1.8 BTC due to ColdCard https://bt.irlbtc.com/view/5589972.0

They confirmed that this compromise was possible due to low-entropy values generated by the MK3 model, an issue that "is present on every Mk3 firmware version since 4.0.1". In response to this they urgently released new firmware, namely v5.6.0 for MK4 and MK5 as well as v1.5.0Q for Q model. These are all mandatory updates.

It is your hope that I can understand, if I had fallen into such situation like you did, I would have made my terrible action already but please stay strong and fight because you have your family behind to take care of them and surely they are really ready to take care of you, share the pain and loss you have now.

About the attacker(s), I don't know, maybe they visited the forum or will do it in the future, but honestly even I wish the best for you, I don't believe that such people will be ready to do the right things like returning bitcoins they hacked to victims including you. If they were good people, they would not do that already.

Will hackers behind this again come from North Korea like Lazzarus, let's wait for investigation and results later.

If you have a draft of that thread, you can make a new one in another board.
Hardware wallets

If you did not save it on your computer, you can find it in your draft page https://bitcointalk.org/index.php?action=drafts

What I find interesting here is not so much the vulnerability itself, but what this tells us about Bitcoin security in the age of AI.

Bitcoin itself is probably one of the most reviewed pieces of open-source software in the world. But the Bitcoin ecosystem is much bigger than Bitcoin Core. Wallets, hardware devices, firmware, libraries, exchanges, signing tools… there is a huge amount of code around Bitcoin, and obviously not all of it has received the same level of scrutiny.

And I think AI changes the game quite a bit here.

Until recently, going through a large and unfamiliar codebase looking for some obscure weakness required a lot of time and some pretty specialised skills. It still does, of course, but AI can make that process much faster. You can analyse more code, follow dependencies, spot strange patterns and test ideas at a scale that simply wasn’t realistic before.

What worries me most is old code.

We tend to think that if something has been around for 5 or 10 years without being hacked, it must be reasonably safe. But that’s not necessarily true. Maybe the vulnerability was always there and it simply wasn’t worth the enormous amount of work required to find it.

AI changes that calculation.

Of course, defenders have access to exactly the same tools, and hopefully this means we will see much more AI-assisted auditing of wallets, firmware and other Bitcoin infrastructure.

But attackers only need to find one forgotten piece of code protecting enough money to make the effort worthwhile.

So I don’t think the real question is whether AI can somehow “break Bitcoin”. That’s probably the wrong way to look at it.

The more interesting question is how much vulnerable code is still sitting somewhere in the Bitcoin ecosystem, considered safe mainly because nobody has managed to find the bug yet.

My guess is that over the next few years we’re going to find out. In a brutal way, probably.

Polymerbit was made by italian and uses keys generated offline and printed on a sort of polymer banknote with a holographic layer. I don’t see any issues if you trust the people who created them, if you trusted them before, you can continue to do so.

It’s very likely that whoever carried out this attack used advanced AI agent to scrape the Coldcard firmware, found the bug, and trained another agent to carry out the attack. So now we should expect all wallets to be combed through by agents created by malicious hackers to find other bugs.

In the coming weeks, there will be a flood of requests for updates to existing wallets, patches, upgrades, phone calls, emails… pay close attention.

https://x.com/COLDCARDwallet/status/2083155034762621425

Coldcard posted this on their official page on X.com. comments surrounding this tweet suggests they should send this to the emails of those who have bought from them others are saying they should refund the victims and shut down.

My heart goes to all those who are victims of this. Please stay strong.

A quick discovery. COLDCARD has also prepared a migration guide to a safer one. It's best to do this as soon as possible or move to another wallet first. Don't use COLDCARD temporarily. Although they say Mk4, Q, and Mk5 won't affect it, it could be risky.

I've never encountered a problem like this before. My question is, will COLDCARD cover the losses?

The attacker has swept addresses, not UTXO, holding more than 0.15 BTC.
As had an address-balance index, not a UTXO set, aggregating balances per address requires a separate indexing layer: an Electrum node, a blockchain API, or a public dataset dump. That's a real infrastructure commitment (a full address index is on the order of a terabyte and takes days to sync), or a third-party dependency (which is very improbable for obvious opsec reasons).

The target list was built in batches well before execution: you don't aggregate address balances across the chain within a 15-minute window.

A strange consequence is that taking the small outputs inside the address that had accumulated 200 UTXOs costs him a fee and blockspace for negligible value. He did it anyway because his unit of work was the address. That's deliberate engineering, or bad engineering.

An idea is that 0.15 may not be a chosen parameter at all. If he sorted candidates by balance descending and truncated at the top N, the floor is simply wherever the last one landed. The sweep ran across exactly 500 transactions, a very round number. If the count is 500, the binding constraint was target count, not index size, and 0.15 is simply a consequence.

The remaining addresses are not safe. Suspect the attacker could launch a second wave of attacks, targeting the remaining unsecured balances. He surely already has all the keys; he just needs to run the script again.

To me, this screams AI engineered tools to swipe the balances.

Your words are worrying, and probably reading them elsewhere would not have made too much of an impression on me, but reading them here I think it is something that we should all take into account and try to improve, as much as possible, the security of our funds. But, in cases where it is very difficult to foresee that a vulnerability would be found, how do you protect yourself?