Good morning, everybody.
It’s your Daily Crypto News for Monday, October 12th, 2026. My name is Matt.
And this is my sixth take this morning.
Matt didn’t sleep last night.
My brain is fuzzy, my mouth is mushy, and my eyes are blurry. I went to bed around 11, probably fell asleep around 11:30, and woke up at four o’clock in the morning. Then I spent another hour and a half scrolling through my phone instead of going back to sleep.
I had a couple of beers last night, they didn’t sit well with me, and now here we are.
So bear with me this morning because I’m not recording this again.
Before we get into the news, please leave a review on Apple Podcasts or Spotify. You can also send me an email at matt@dailycryptonews.net.
And a couple of quick shout-outs.
SRT Scat left a comment saying the show has become part of their routine and helps them stay informed while they’re on the go. Thank you very much.
And underscore Caitlin sent a shout-out from Calgary, Canada, saying they’re here for the rants and insights.
Well, you’re definitely getting a rant today.
Because I spent a good portion of the weekend going down a rabbit hole about artificial intelligence, mathematics and the possibility that AI could eventually threaten the cryptography securing Bitcoin, Ethereum and much of the modern internet.
And I have to tell you, I’m taking this a lot more seriously than I was last week.
Bitcoin Holds $83K, but the Bigger Story Is AI and Cryptography
Bitcoin is holding around $83K to $84K this morning after testing $80,000 last week.
We didn’t break through $80K.
That’s good.
But I still think we’re going to test that level again, and I wouldn’t be surprised if we eventually break below it.
I don’t mean to be Mr. FUD here.
We’ve been watching this same trading channel for weeks. Bitcoin needs to break above $87K, establish something around $84K to $86K as support, and then start targeting $90K.
Until that happens, we’re still basically going sideways.
But Bitcoin’s price isn’t what I want to spend most of today discussing.
Over the weekend, I was listening to the All-In Podcast, as I usually do.
And to be perfectly honest, I’ve been tuning out a lot of their AI conversations lately.
It feels like they’ve been talking about nothing but AI for six months.
At some point, it all starts sounding repetitive.
But this weekend’s conversation was different.
David Friedberg was discussing a major collection of mathematical research released by OpenAI, and it connected directly to something we’ve been talking about here on Daily Crypto News.
Last week, we discussed Ethereum researcher Justin Drake’s warning that advances in artificial intelligence could eventually threaten the mathematics protecting cryptocurrency wallets.
At the time, I didn’t fully appreciate what was happening.
I thought a lot of the controversy was about mathematicians wanting recognition for their work and people arguing about whether AI should receive credit for solving mathematical problems.
That was part of the conversation.
But after listening to the discussion and digging into the underlying cryptography, I realized there was a much bigger question.
What happens if AI discovers new mathematics that makes supposedly impossible cryptographic problems much easier to solve?
And this isn’t just a Bitcoin problem.
This could affect Ethereum, banking, secure communications, authentication systems and enormous portions of the digital infrastructure we use every day.
OpenAI’s Mathematical Research Raises a New Security Question
On October 6th, OpenAI released a large collection of mathematical research generated by internal frontier AI models.
These are research systems beyond the models most people interact with publicly.
According to the figures discussed, the models were given approximately 4,000 open mathematical research problems.
The initial release included around 722 mathematical manuscripts, organized into 372 families of results.
The work covered subjects including number theory, algebra, geometry and theoretical computer science.
OpenAI also released machine-checkable formulations for many results using Lean, a system that allows mathematical proofs to be formally checked.
Think about that for a minute.
We’re talking about AI systems producing hundreds of mathematical research manuscripts that professional mathematicians are now evaluating.
And the models doing this work aren’t necessarily the same ones available to everybody using ChatGPT.
That opens up a completely different conversation about what frontier AI systems might already be capable of.
I’ve said before that I think artificial superintelligence could arrive much sooner than people expect. I even wonder whether some internal systems are already substantially more capable than the public realizes.
That’s speculation on my part.
But we’re sitting here arguing with ChatGPT about not using dashes while research models are producing mathematical work that has people in cryptography asking serious questions.
Now, here’s why this matters.
Modern cryptography depends on mathematical problems that are believed to be extraordinarily difficult to solve.
The security doesn’t necessarily come from the mathematics being secret.
It comes from the enormous difference between performing a calculation in one direction and reversing it.
And if AI discovers a new mathematical shortcut, some of those assumptions could change.
Why Bitcoin’s Private Keys Are So Difficult to Recover
Bitcoin and Ethereum both rely heavily on elliptic-curve cryptography.
The basic idea is relatively straightforward.
You have a private key, which is essentially a secret number.
Using that private key, your wallet can calculate a corresponding public key.
That’s easy.
But going backward is an entirely different problem.
Given a public key, calculating the private key is believed to be computationally infeasible using today’s conventional computers and known algorithms.
The underlying mathematical challenge is called the elliptic-curve discrete logarithm problem.
And for decades, mathematicians and cryptographers have studied these problems without discovering an efficient classical algorithm capable of solving them at the scale used by modern cryptocurrency systems.
That’s an important distinction.
We haven’t mathematically proven that no efficient algorithm could ever exist.
We simply haven’t found one.
And that’s where artificial intelligence enters the conversation.
AI doesn’t necessarily need to guess trillions upon trillions of private keys faster than a normal computer.
That isn’t the threat Justin Drake is describing.
The threat is that AI might discover a fundamentally better way to solve the mathematics.
Imagine that the best-known attack against a cryptographic system requires an impossibly enormous amount of computation.
Now imagine AI discovers a mathematical transformation that reduces that computational requirement by several orders of magnitude.
That’s different from building a faster computer.
You’ve changed the problem itself.
And if the improvement is significant enough, something previously considered impossible could potentially become practical.
Drake has warned that AI-driven mathematical discoveries could accelerate the timeline for these kinds of threats much faster than previously expected, potentially even into a matter of months in an extreme scenario.
Other cryptographers strongly disagree with that timeline.
And I think that’s important to acknowledge.
Nobody has demonstrated that AI can currently recover Bitcoin private keys from public keys.
Elliptic-curve cryptography has not been broken.
But the possibility of new mathematical discoveries is becoming a more serious subject of discussion.
And after spending the weekend looking into this, I understand why.
The AI Research Wasn’t Perfect
There’s another part of the OpenAI story worth discussing.
Not everything the models produced was correct.
On October 7th, OpenAI reportedly withdrew three manuscripts after a sign error invalidated part of a proof and two dependent results.
Another 14 manuscripts received corrections or repairs.
That’s important.
AI can produce sophisticated mathematical work and still make mistakes.
We shouldn’t automatically assume every generated proof is valid simply because a frontier model produced it.
But the errors don’t make the broader research meaningless.
The original release contained hundreds of manuscripts, and professional mathematicians were evaluating the results.
That’s still remarkable.
The important question isn’t whether AI is suddenly perfect at mathematics.
It’s whether AI is becoming capable of helping researchers discover things that would previously have taken substantially longer.
And if it can do that, we have to start thinking seriously about what happens when it turns its attention toward cryptography.
Three Ways AI Could Change Cryptocurrency Security
I think there are three broad possibilities.
Scenario One: Nothing catastrophic happens.
AI becomes an increasingly powerful mathematical research tool.
It helps mathematicians solve difficult problems, develops new proofs and improves existing algorithms.
But it never discovers an efficient way to break elliptic-curve cryptography.
Bitcoin and Ethereum continue using secure cryptographic systems, and the industry benefits from better research.
This is entirely possible.
Scenario Two: AI improves existing attacks without making them practical.
This is probably the more realistic concern.
Imagine that recovering a private key currently requires an absurdly enormous amount of computation.
AI discovers a mathematical improvement that makes an attack ten times faster.
Or a thousand times faster.
Or even a billion times faster.
That sounds terrifying.
But if the original problem would have required an unimaginably long period of computation, a billionfold improvement might still leave the attack completely impractical.
That’s something people need to understand.
A dramatic improvement in an algorithm does not automatically mean a cryptographic system has been broken.
The question is whether the improvement is large enough to make an actual attack feasible.
Scenario Three: AI discovers a fundamental mathematical shortcut.
This is the nightmare scenario.
Imagine an AI system discovers an efficient classical algorithm for solving the elliptic-curve discrete logarithm problem.
If that happened, an attacker might be able to derive private keys from sufficiently exposed public keys without waiting for a powerful quantum computer.
That’s the scenario that could change everything.
And not just for cryptocurrency.
We’re talking about financial institutions, government systems, secure communications and digital authentication infrastructure.
Much of modern internet security relies on cryptographic assumptions that could be affected by a breakthrough of that magnitude.
Now, would the entire internet collapse overnight?
Not necessarily.
Different systems use different cryptographic algorithms, and many could migrate to alternatives.
But the transition could be enormously disruptive.
And if a single organization discovered such a vulnerability before anybody else, the implications could be extraordinary.
Which brings me to another thought I’ve been having.
What If the Most Powerful Weapon Isn’t Controlled by a Country?
We’ve spent generations thinking about national security in terms of countries.
Russia.
China.
Iran.
The United States.
Missiles, nuclear weapons, aircraft carriers, armies and all the other things countries build to defend themselves or project power.
But what happens if the most powerful technological capability on Earth is developed by a private company?
What if a company develops an AI system capable of exploiting critical digital infrastructure on a scale that governments cannot easily counter?
I’m not saying that’s happened.
I’m not saying Sam Altman or Dario Amodei is sitting in an office somewhere controlling the world’s banking systems.
But think about the implications.
Historically, a company trying to challenge a government would face enormous financial and military limitations.
Countries have armies.
Countries have resources.
Countries have the ability to mobilize entire economies.
But if a sufficiently advanced AI system became capable of exploiting critical infrastructure, the balance of power could look very different.
Maybe the future security threat isn’t always another country.
Maybe it’s a private organization controlling technology that nobody else understands or can effectively counter.
That’s a pretty wild thought.
And it’s one of the reasons I think we need to understand what frontier AI systems are actually capable of doing.
AI and Quantum Computing Are Two Different Threats
There’s another development that makes this conversation interesting.
Researchers participating in the Elliptic Curve Fail project recently worked on improving a component of a hypothetical quantum attack against elliptic-curve cryptography.
More than 100 participants reportedly worked alongside AI coding agents, producing improvements that reduced estimates for an important quantum-circuit component by more than 50% compared with an earlier benchmark.
That does not mean the total computing power needed to attack Bitcoin has suddenly been cut in half.
It means researchers found a substantial improvement in one component of the proposed attack.
But that’s still meaningful.
It demonstrates that the algorithms themselves can improve, not just the hardware.
And AI can help accelerate that process.
Here’s the distinction I think people need to understand.
Quantum computing: We already know a mathematical algorithm that could threaten elliptic-curve cryptography. It’s called Shor’s algorithm.
The problem is that we don’t yet have a sufficiently powerful, fault-tolerant quantum computer to execute that attack against the cryptographic keys used by Bitcoin and Ethereum.
We know the general mathematical approach.
We don’t have the necessary machine.
Artificial intelligence: We already have powerful computers capable of running advanced AI models.
What we don’t know is whether an efficient classical algorithm exists that could make recovering private keys practical.
AI might help researchers discover one.
Or it might never find one because no such useful shortcut exists.
That’s the difference.
With quantum computing, we know the algorithm and are waiting for sufficiently capable hardware.
With AI-assisted classical cryptanalysis, we have the hardware and are asking whether a radically better algorithm can be discovered.
And both possibilities deserve attention.
Why Some Bitcoin Wallets Could Be More Exposed Than Others
Let’s talk about what this could mean for people actually holding Bitcoin.
An attacker attempting to recover a private key through a hypothetical elliptic-curve attack would generally need access to the corresponding public key.
Depending on the address type and how a wallet has been used, Bitcoin can sometimes keep that public key hidden until coins are spent.
Once certain transactions reveal the public key, that information becomes available on the blockchain.
Ethereum has a similar distinction.
An ordinary Ethereum account that has never signed a transaction generally has not directly revealed its full public key onchain.
Once it begins signing transactions, the public key can be reconstructed from the signatures.
And that creates an interesting security consideration.
Justin Drake has suggested that large, sophisticated holders consider what he calls “bunker mode.”
The general idea is to minimize unnecessary public-key exposure and consider using fresh addresses whose relevant public keys haven’t previously been revealed.
Now, that isn’t a universal solution.
It doesn’t eliminate every possible cryptographic vulnerability.
And not every Bitcoin address format has the same exposure characteristics.
But it’s an example of how wallet design and operational security could matter if future attacks become more capable.
Vitalik Buterin has also taken a measured position.
He agrees that AI-accelerated mathematical research deserves serious consideration, but he has warned against people panicking and frantically moving their cryptocurrency.
And I agree with that.
Because moving cryptocurrency introduces risks of its own.
Anybody who has sent a substantial amount of Bitcoin knows exactly what I’m talking about.
You copy the address.
You double-check it.
Maybe you send a small test transaction first.
Then you send the larger amount.
And you sit there waiting for confirmation.
That moment between pressing send and seeing the transaction arrive is stressful as hell.
It doesn’t matter whether you’re sending $1,000, $10,000 or $10 million.
If you make a mistake, you may never get that money back.
So I’m not suggesting everybody immediately start moving their funds because of a hypothetical AI threat.
I’m saying the industry needs to take the research seriously and prepare appropriate solutions.
What Happens If Elliptic-Curve Cryptography Actually Breaks?
Bitcoin could theoretically migrate to a different signature system.
Ethereum could do the same.
But think about what that would involve.
Protocol changes.
Wallet software updates.
Hardware-wallet upgrades.
Exchange migrations.
Custodian migrations.
Users moving funds.
And potentially enormous amounts of cryptocurrency needing to be transferred into new security arrangements.
Then there are dormant wallets.
Lost Bitcoin.
Abandoned accounts.
People who haven’t touched their coins in ten years.
And, of course, Satoshi’s Bitcoin.
What happens to assets controlled by keys that might eventually become vulnerable but whose owners cannot or will not migrate them?
That’s a complicated problem.
It would require serious technical decisions, coordination and potentially contentious debates about how different blockchain networks should respond.
Which is another reason to think about these issues before there’s an emergency.
The Biggest AI Security Threat Might Still Be You
Here’s something else.
AI doesn’t actually need to break elliptic-curve cryptography to steal cryptocurrency.
It could become extremely good at finding software vulnerabilities.
Smart-contract bugs.
Wallet implementation errors.
Exchange security weaknesses.
Problems in applications that interact with blockchains.
Or vulnerabilities that have existed for years without anybody noticing.
And then there’s what I think might be the biggest weakness of all.
Human beings.
I’ve been spending a lot of time experimenting with AI-generated images and videos.
Over the weekend, I went down another rabbit hole looking at AI filmmaking tools, including Higgsfield.
The technology is getting unbelievably good.
We’re talking about AI-generated characters, voices, music, visual effects and entire cinematic sequences.
Some of this stuff is amazing.
But think about what that means for fraud.
Imagine receiving a video call that looks and sounds exactly like somebody you know.
Your mother.
Your brother.
Your spouse.
Your boss.
Imagine getting an email that perfectly imitates the writing style of somebody you trust.
Or a text message that seems completely authentic because the AI knows enough about your relationships and communication habits to make it convincing.
You don’t need to crack a Bitcoin private key if you can convince the owner to give you access.
You don’t need to break a blockchain if you can manipulate somebody into signing a malicious transaction.
And AI could make those attacks much more sophisticated.
That’s social engineering.
And we’ve been dealing with it for decades.
AI just gives attackers a much more powerful set of tools.
That’s why I think the conversation about AI and cryptocurrency security needs to go beyond whether a particular mathematical algorithm is vulnerable.
We have to consider the entire security environment.
The cryptography.
The software.
The wallets.
The exchanges.
And the people using them.
Because ultimately, the weakest point in the system might not be the mathematics.
It might be us.
Crypto Prices
All right, that was a lot.
This was supposed to be a relatively short podcast, and somehow I’ve spent almost half an hour talking about elliptic curves, quantum computers and AI potentially taking over the world.
Let’s get into prices.
Bitcoin is sitting around $83,350, up approximately 0.5% over 24 hours.
Ethereum is around $2,150, up approximately 2%.
USDT remains number three.
BNB is trading around $749, up approximately 0.3%.
XRP is approximately $1.40, up around 0.9%.
USDC remains number six.
Solana is around $111, up approximately 1.8%.
TRON is sitting around $0.331, up approximately 0.3%.
Zcash is around $1,300, up approximately 6%.
Hyperliquid is approximately $86.22, up around 1.4%.
Litecoin is around $64.82, up approximately 2.2%.
Total Crypto Market Cap: approximately $2.83 trillion, up around 0.9%.
Fear & Greed Index: 57, Neutral.
My Take
I started last week thinking the discussion about AI breaking cryptography was mostly another round of speculation.
And I was wrong to dismiss it that quickly.
After listening to the All-In Podcast and digging further into the mathematics, I think this deserves a much more serious conversation.
Not because Bitcoin has been broken.
It hasn’t.
Not because AI has suddenly discovered how to recover private keys.
There’s no demonstrated evidence of that.
But because the entire security model depends on mathematical assumptions that researchers have spent decades studying.
And now we’re developing AI systems that can potentially accelerate mathematical research at an extraordinary rate.
That’s worth paying attention to.
I think the most important distinction is that AI doesn’t need to become infinitely powerful at brute-force computation.
It needs to discover better mathematics.
And if it does, the implications could extend far beyond Bitcoin.
We’re talking about modern digital infrastructure.
Banking.
Communications.
Government systems.
And enormous amounts of value protected by cryptographic keys.
The worst-case scenario is frightening, but it also isn’t inevitable.
Maybe AI never discovers a practical shortcut.
Maybe the research leads to better cryptography instead.
Maybe the industry uses these advances to strengthen security long before any serious threat emerges.
That’s what I hope happens.
And I think it’s why we should be encouraging this research rather than dismissing it as FUD.
There’s also a bigger conversation about who controls these increasingly powerful AI systems.
I find it fascinating that we’re developing technologies capable of performing advanced mathematical research while most people have very little visibility into what the most capable internal models can actually do.
I don’t know whether artificial superintelligence already exists somewhere behind closed doors.
I suspect the frontier of AI research is much further ahead than most people realize, but that’s my speculation.
What I do know is that the capabilities we’re seeing publicly are already extraordinary.
And we need to start thinking seriously about what happens when those capabilities affect critical infrastructure.
As for Bitcoin’s price, we’re still around $83K.
We held $80K last week, but I still think another test is possible.
We need to reclaim $87K and eventually start using the mid-$80,000 range as support before I’m going to get particularly excited about another run toward $90K.
For now, I’m watching the price, watching the AI research and trying to get a better understanding of where all this technology is going.
And I’m also going to try to get some sleep tonight.
This was supposed to be a short show.
It was not.
I’m still waiting for Craig Cobb to send over his market update, so you may get another Daily Crypto News episode this afternoon.
Have a great Monday.