Verifying bitcoin code

[trel]

Hello, has there been an initiative for verifying all of the bitcoin core's code or other implementations of the bitcoin protocol (btc knots, bitcoin in Go.. etc) with the help of Claude and especially the models Fable or Mythos? I know the codebase is big and the task won't be just straight forward but there are ways to achieve this, possibly privately first and maybe tokens paid by a bitcoin grant(not sure how that works for an open source protocol.. maybe there are people with big enough investments willing to sacrifice a couple of $ for this verification)

[Cookdata]

There are some organizations that support and compensate Bitcoin developers, this shouldn't be a problem if they have been supportive all these years, maybe they are doing it already.

Who knows if there are some group doing independent review with AI for some exploit just like the recent one found on zcash but Bitcoin bugs are rare because proposals takes time, some take years before they are implemented to avoid problems later.

[ABCbits]

I never heard about such initiative. I did quick search on https://groups.google.com/g/bitcoindev/ and opensats.org, but there's no relevant result (unless i missed something).

especially the models Fable or Mythos?

The one suspended by US government?

maybe there are people with big enough investments willing to sacrifice a couple of $ for this verification)

Who knows if there are some group doing independent review with AI for some exploit just like the recent one found on zcash

FWIW, the one who found critical vulnerability on zcash[1] is a security consultant paid by shielded labs[2]. So cost of both human expert and the AI will be far more than couple of dollars.

[1] https://forum.zcashcommunity.com/t/the-orchard-counterfeiting-vulnerability-and-next-steps/56015
[2] https://forum.zcashcommunity.com/t/shielded-labs-engages-taylor-hornby-as-security-consultant/55421

[trel]

Fable is now suspended, yes ( I think it will be temporarily). Opus 4.8 would still be a good option but Fable/Mythos are best for this as of now. I know it will cost some amount of money, not sure how much, but as mentioned there are grants that can be used for this. Sounds better than the alternative which is someone finding an exploit that's very hard to catch by a human eye and possibly hurting the protocol or causing a lot of fear and retreat from BTC (some bugs went unnoticed for many years and were found by Mythos recently, two articles that come to mind https://rival.security/posts/mythos-discovered-a-cve-already-in-its-training-data---and-thats-still-worrying and https://arstechnica.com/security/2026/05/google-publishes-exploit-code-threatening-millions-of-chromium-users/ but there are more I believe). My point being - I think it's a good direction to push and spend some grant money for.