The bigger disaster is awaiting. hackers are now reviewing source code of everything crypto related for entropy or other misconfiguration bugs, automated with leading AI models , which will spot errors humans missed. We're talking wallets dating back to 2010 when btc was founded, and then for all the altcoins. Ppl don't realize how bad this will get when exploit detection can be done at scale with LLMs. Ai can review a codebase in minutes, which would take humans days, if ever. Coldcard is a tiny wallet service. A bigger one would have losses in excess of billions. A hacker will first find the vulnerability and then exhaust the keyspace and download the blockchain to cue up the keys and do it all at once .
back
1 comments
Better now than later. It would have been even better a decade ago, but here we are.
Will this just trigger a crypto arms race? Bigger and tougher proof of work?
Unlikely. The digital signatures bitcoin uses are regarded as secure, (there could be attacks if we get quantum computers.) The issue here was the software didn't put in enough randomness when generating the keys, and so it was straight forward to search the small space of keys the wallet would generate.
It's not so different if you encrypted a file with a weak password and someone guessed it—you don't need stronger encryption, you need to pick a better password.
The proof of work part of bitcoin is really for "consensus", so everyone agrees which transactions happened in which order, this attack didn't affect that part at all.
The issue is they forked sane GPL code (Trezor's), then decided they wanted no one to be able to copy them in turn so they rewrote components and messed up entropy. IIRC they intended to still use the secure element to generate some but it failed open and had no meaningful tests to detect so, leading to significantly fewer bytes for years.
Tests should have caught that, and it should not have simply failed open during the most critical part of entropy generation.
Claude has found some weaknesses in other schemes, but not fully breaking anything yet:
https://www.anthropic.com/research/discovering-cryptographic...
I am always hopeful that something like this might kill cryptocurrency (although I don't believe it will) - if a critical mass lose confidence in it, the network effect would collapse and people might stop wasting terawatt-hours of power on NFTs and exit scams.
why do you care so much about that? NFTs are basically dead already and crypto is still somewhat niche
like, why does it even matter? I doubt the energy usage is anywhere near AI workloads running today
Setting aside any utility it might have, crypto has lots of negative externalities besides energy usage. The event under discussion, for example.
In what way is energy usage in creating money a negative?
e.g.
If money took no energy to create, what would it be worth? What if you limited its effortless creation to just a few people?
What do you think about the enormous energy usage (far more than bitcoin) in mining gold?
This has nothing to do with proof of work
Need more GPUs.