i put this together. it also contains a mode for CVE-2023-33297 ('headers') which was recently patched - although i wasn't given credit because i have haters @ bitcoin-core and blockstream. they kinda run the show now. i'm even banned from contributing to the bitcoin-core github. much decentralization.
that's... not at all a solution. "just use onfido" doesn't solve deepfakes. too much friction isn't true whatsoever it would just take some solid coders. and finally i don't suggest that anyone work with onfido. they fleeced me out of a $10k deposit in 2018 for kyc services that were falsely advertised as deeply international only for us to realize that wasn't the case at all. no refund. no partial refund. no access to credits anymore. $10k gone because onfido is shady.
i came up with this. it has a lot of moving parts but it's viable. i understand the idea of "why not just publish transcripts in advance on whitehouse.gov" and the answer is because there are too many points of failure, like the risk of shells, rogue actors, dns hijacking, ddos, etc.
these issues are alleviated by multiple blockchains running byzantine fault tolerance against speeches that are monitored and authenticated by public facing figures both beforehand and during the live broadcast.
i know it seems complicated - but it needs to be.
happy to answer any questions if they aren't rude. i understand that this is a lofty proposal and the natural reaction is a knee-jerk "yeah ok" - but give it a read. play with it in chatgpt. this is about viability and computer science theory more so than a tangible effort in establishing a funded operation.
tl;dr blending bleeding edge technologies together to 360 noscope deepfakes.
>right, because so many people are typing in the incorrectly shortened URL instead of clicking on the link
it was clickable - and yes they dodged a bullet. the channel could have been perfectly reproduced, made to look real, and then used to intercept western intelligence. there was no manual "typing" of links
was me if anyone wants details - but it's pretty simple. not a hack. registered an available username so that opposition ic groups couldn't do the same - they could have dressed it up in a visually identical way and intercepted western intel as a best case scenario.
it's just an intelligence failure. those aren't new - nor is this really even that big of a deal - imo - it just could have hypothetically been a big deal - and i don't really care for hypotheticals
i discovered the issue if anyone would like for me to unpack anything.
content snippet + attack code:
the takeaway from pad vs bitcoin is that a botnet can remotely charge public listening node operators thousands of dollars.
that would snowball into a mass exodus of public nodes until all that remained were millions lost unexpectedly by thousands of node operators - and a network centralized by entities able to weather the cat 5 hurricane of upstream overage fees, like:
1. mining pools
2. exchanges
3. hnwis
4. governments
effectively centralizing the network and causing countless millions in damages. the ethos of bitcoin and the financial survival of its operators are at stake.
i put this together. it also contains a mode for CVE-2023-33297 ('headers') which was recently patched - although i wasn't given credit because i have haters @ bitcoin-core and blockstream. they kinda run the show now. i'm even banned from contributing to the bitcoin-core github. much decentralization.
how to reproduce the vulnerability yourself:
1. install bitcoin
2. ./bitcoind
3. edit your ip into attack.go: https://pastebin.com/raw/qr8XWFkR
4. save attack.go
5. snap install go --classic
6. go build attack.go
7. ./attack
8. slow oom crash - enjoy the fireworks
screenshot: https://i.imgur.com/DA80ORS.png
happy to answer any questions