Zcash 经受 AI 破解考验后创历史新高

In May [music] 2026, a security researcher turned a frontier AI model loose on [music] Zcash. And in just under a day, it found a bug that 4 years of live operation and multiple cryptographic [music] audits had overlooked. Now, many assumed that was a death sentence [music] for Zec. The idea was that privacy coins were a pre-AI idea. Machines cannot break them, [music] and this is what the end of this trade actually looks like.

But 3 months later, Zcash is now trading at a new all-time high of $800 $888, [music] up 80 plus percent in a single month. And the bug that was supposed to kill Zcash ended up showing everybody why the protocol was built this way in [music] the first place. So, today, we're breaking down what the AI actually found, why the network survived something that should have been fatal, and why AI is now arguably the single biggest tailwind [music] this asset has.

My name is DC, and this is The Coin Bureau. So, let's start with the bug that was supposed to spell the end of Zcash. This was not a sloppy smart contract, and it wasn't an exchange getting fished. The issue with Zcash was a matter of cryptography. Specifically, the bug was in the cryptographic code behind Zcash's Orchard shielded transactions, and it had been live on the main network since network upgrade five in May 2022.

That's 4 years in production. The researcher was Taylor Hornby, working on a contract with Shielded Labs, and the discovery was credited to an AI auditing harness built around Anthropic's Claude Opus 4.8. Roughly 6 hours of targeted analysis, and then a working local proof of concept. Now, what actually was it? In simple terms, it was possible to create a zero-knowledge proof that passed all of Zcash's checks, even though the transaction it was supposedly proving wasn't actually valid.

It was a bit like a bouncer who inspects every single ID with perfect precision, checks the hologram, checks the font, checks the expiry date, but never actually checks whether the person on the card actually exists. And in a shielded pool where amounts and participants are hidden by design, that means conjuring notes that were never supposed to exist with no public trace at all. In a local test environment, it was confirmed capable of producing unlimited counterfeit Zec.

Not a nice sentence to read if you're a holder. And the reason this was important well beyond crypto is that a single researcher with the right AI tooling did in about a day what a well-funded multi-audit human review process had not managed in four years. So, every bank and every code base out there started looking much more vulnerable to AI. Zcash just happened to be the first name on the list. So then, how the hell is this coin at new all-time highs rather than at zero?

Well, there are three major reasons. The first is that Zcash was actually designed with this kind of failure in mind. Its developers assumed that one day even their own cryptography might break, so they built a safe guard called the turnstile. The turnstile basically keeps track of money moving between different parts of the Zcash system, such as from transparent transactions into shielded ones or from an older shielded pool into a newer one.

The rule is that more money can come out of a pool than has been verified going into it. Individual shielded transactions can remain private, but the total amount entering and leaving each pool is still publicly tracked. So, even if the privacy system itself were completely broken, an attacker secretly create extra Zcash beyond the 21 million cap and then withdraw it unnoticed. Any money leaving the pool still has to pass through this publicly counted checkpoint.

That is a system designed by people who assume something could eventually go wrong. And when you're designing money, that kind of paranoia is a feature. The second reason is the response, which was incredibly fast. Horbit disclosed privately the same evening he found the bug. Within roughly 72 hours an emergency soft fork shipped by a Zcash 4.5.3 disabling Orchard shielded transactions network wide to remove the attack surface before the vulnerability was public.

On the 3rd of June an emergency hard fork with a corrected circuit restoring shielded functionality. That's only the second security driven protocol upgrade in Zcash's entire history. Engineers at the Zcash Open Development Lab, Daira Emma Hopwood, Chris Nuki Khum and Jack Krieg among them confirmed and patched within hours. Now compare that response to the industry norm and the difference is stark. And the third reason Zcash could recover from this stress test is well, a little strange.

Nobody could prove the pool was clean. Because Orchard is quite literally private, there's no forensic method to demonstrate that the flaw was never exploited during those four years. Shielded Labs assessed prior exploitation is unlikely given the complexity involved and the pool's balance had grown steadily with no suspicious outflows, but they did acknowledge that you can't prove a negative here. That's quite the admission from a project while its price was collapsing and collapse it did.

Zcash collapsed by 50% with over $5 million in market cap wiped out at the trough and several high profile holders liquidating publicly. Ouch. So, what did the developers do with an unprovable question hanging over $1.7 billion of shielded value? Well, they didn't patch and pray. On the 28th of July the Ironwood upgrade activated and rather than keep the repaired Orchard pool running, it retired Orchard permanently. The old pool was effectively locked into exit only mode behind the turnstile.

That means any fake Zcash that might have been created there during those four years is now trapped inside and can never all be withdrawn. It was replaced by a brand new shielded pool built with a clean code. According to one account, that new system was tested against more than 2,700 mathematical proofs using automated tools before it ever went live. So, in other words, the system was formally verified and users quickly moved over.

Around 176,000 Zec entered the new Iron Wood pool on day one and by mid-August it had already become the largest shielded pool on the network. So, Zcash never proved that the old bug was never exploited, but it did make sure that any exploit from the old system was no longer relevant.

What if you could trade real US stocks like Apple, Nvidia, or Tesla without leaving your crypto account? Well, that's the idea behind our tokens from Bitget. These are tokenized stocks backed one-to-one by real shares, but the key difference is they are actually usable. You can trade them, use them as margin, and even earn dividends instead of just letting them sit in your wallet. So, if you want to check them out for yourself, sign up for Bitget using the link in the description or by scanning this QR code.

Which brings me to an interesting point, if I do say so myself, because plenty of protocols have survived a scare, but almost none of them survived an existential scare of this magnitude. I mean, let's be honest, most cryptos barely survive at all. Go back to the coins that were household names in 2017 and 2018 and look at where they trade as I make this video. EOS is 99% below its 2018 peak. NEO is down 99%. Dash, once a genuine top 10 asset, is down 97%.

Litecoin is 87% below its December 2017 high. Ethereum Classic is down 83%. Cardano, 83%. Stellar, 79%. And XRP, the strongest survivor of the lot, is still 61% below where it traded in January 2018. Cycle after cycle and still nothing. So, why do old altcoins stay dead? Well, there four main reasons and none of them are about the tech. First, venture unlocks that simply never stop hitting the market. Second, perpetual emissions which create a class of eager sellers.

Third, developers mindshare draining away toward whatever is newer. And fourth, the marginal buyer of this cycle is an institution with an ETF wrapper and a compliance depa