CYBERATTACKS, BANKING RESILIENCE, AND BITCOIN’S ROLE IN A CRISIS
Payment vulnerabilities, bank defenses, artificial intelligence, and the limits of digital assets in a systemic collapse
A cyberattack could bring some payments to a standstill without destroying the banking system. Resilience depends on multiple payment rails, strong safeguards, and cash on hand. Bitcoin can serve as an alternative in a crisis only if you hold it in self-custody and are prepared to use it.

☣️ Crédit Mutuel Arkéa, Binance, Ledger, and refrigeration systems at 700 supermarkets. In just a few weeks, cyberattacks came one after another. Could an attack stop us from making payments? And what should you do if your bank stops responding?
📺 Jonathan Herscovici from Surfin’ Bitcoin recently interviewed me about how banks defend themselves and what we can do to protect ourselves.
🏧 The banking system and #payments infrastructure are now in a state of constant cyberwarfare, with attacks targeting banks, payment institutions, crypto platforms, and public institutions.
⚠️ A major attack could disrupt card payments, bank transfers, cash withdrawals, and access to online services. Resilience depends on keeping systems separate, using multiple payment rails, planning for recovery, monitoring transactions, and having insurance and guarantee funds in place.
💶 It’s prudent to keep enough cash on hand to cover several days of expenses, or even a month’s worth.
🚨 #AI makes it faster to find and exploit vulnerabilities and easier to carry out deepfake attacks. That makes it essential to stay alert and verify suspicious requests through a separate channel.
💡 #Bitcoin can serve as an alternative means of payment in a crisis, provided you hold your funds in self-custody and are ready to use them.
🚫 #ETFs and #stablecoins don’t guarantee direct access to liquidity if your bank account is blocked.
Watch the full video (in french)
Introduction
:: Jonathan :: Hello everyone, and welcome to another episode of the Surfin’ Bitcoin podcast. I’m very happy to be back with you, and it’s a pleasure to welcome back a familiar face on the channel, Michel Khazzaka.
Hello, how are you?
:: Michel :: Hello, and thanks very much for having me. It’s a pleasure to be here with you and everyone listening.
:: Jonathan :: Today we’re going to look at a subject that’s been in the news and imagine what cyberattacks and cybercrime might mean for the banking system. Unfortunately, it’s been making headlines for several months, and I think it’s something you’ll want to know about. Feel free to like and share the video. This is a live issue, and it’s important to understand the risks you face.
Before we get into it, Michel, could you introduce yourself to anyone who doesn’t know you yet?
:::Michel Khazzaka’s background
:: Michel :: Amid all this anxiety about cyberattacks and cybersecurity, let me start with a quick introduction. I’m Michel Khazzaka, a friend of Surfin’ Bitcoin and StackinSat. Several years ago I founded Valuechain, a consulting firm working in payments and crypto.
What does that cover?
- Card payments, bank transfers, and direct debits
- Core banking, fraud, and conventional banking cybersecurity
- Blockchain, especially Bitcoin, and its use in banking and bank transactions
- Stablecoins, the wholesale digital euro, central bank digital currency, and Web3 in banking
:: Jonathan :: Thanks for the introduction. Today we’re going to approach the subject from the payments side, which is your area of expertise. Let’s imagine a world where you can’t pay by card or make a transfer and a very large hack has brought the economy to a standstill.
:::Could a cyberattack bring down the banking system?
:: Jonathan :: Could a cyberattack destroy the banking system and the payment system?
:: Michel :: I’d take issue with the word “destroy,” but otherwise, yes. The risk has never been higher. I’d go so far as to say it’s more than imminent. In banking, we really are at war. It’s a battle between defenders and attackers. There are too many attackers for me to characterize them all here. We’ve just seen attacks involving Crédit Mutuel Arkéa, Binance, and banks that made no public announcement, either because there was fortunately no damage or because the losses were handled through recovery, fraud procedures, or other channels. Either way, the system is under attack.
:::Keeping cash in reserve
:: Michel :: Even the head of France’s card payments sector stressed that people should urgently keep at least a few days’ worth of cash at home in case a large-scale attack hits Visa, the CB network, or Mastercard and stops card payments from working.
What would that mean in practice?
You wouldn’t be able to withdraw cash from an ATM. You’d be unable to pay for your baguette by card or get cash out. Cash withdrawals are part of the card payments infrastructure too. That’s why the recommendation is to keep enough cash at home for several days, or, if possible, the equivalent of a month’s expenses.
Keeping a cash reserve wasn’t common before. Everything was moving toward digital payments. I went cashless myself because it was convenient.
:: Jonathan :: Today, in 2026, paying by card feels like second nature, especially since contactless really took off after COVID. I don’t remember the figures for the share of contactless payments, but I think it’s risen a lot. You start to feel that paying in cash the way our parents did is dying out. I remember buying sweets or bread with coins. Now I pay by card even for a two- or three-euro baguette. I think a lot of people in France have picked up the same habit.
:: Michel :: It’s partly a generational thing. Cash is still widely used in Europe. Less so in France, but in Germany and elsewhere in Europe the average is above 40%, approaching 50% of retail payments. Cash is still very much alive.
:::Contactless payments and Apple Pay
:: Michel :: Paying with your phone or tapping a card is very easy. A phone is even more convenient than a card, unfortunately.
Apple Pay is the clear leader in this field in Europe. If you’ll forgive the expression, it crushes the competition in contactless payments in Europe and in France. We saw that firsthand when I helped banks roll it out. France resisted for too long because it wanted to build its own competitors to Apple Pay. Unfortunately, the figures suggest that effort failed.
:::Wero and instant payments
:: Michel :: Now we have Wero, which gives me a lot of hope. It lets you make an instant bank transfer, say from Michel to Jonathan, using a QR code. A lot of French banks now offer it.
:: Jonathan :: For our listeners, could you explain how Wero works in practice? It’s been hugely successful in recent years. Without getting too technical, how does a simple QR code make paying so easy? What’s going on behind it?
:: Michel :: In our industry we call them payment rails, like railway tracks. They’re different networks and sometimes intersect, but Wero runs on a very different rail from card payments. A bank transfer used to take one to three days, depending on the circumstances. With an instant transfer, the recipient now gets the money in under seven seconds.
An instant transfer still needs an IBAN. It’s cumbersome to ask Jonathan for his IBAN, type in that whole string of characters without making a mistake, and send him €20 for dinner. Banks eventually came together through the European Payments Initiative to make instant transfers easier without asking people to enter an IBAN directly. It’s a private initiative. The system generates a single-use QR code, and the participating bank maps it to the associated IBAN, simplifying that last step for the user.
Cards run on a completely different rail, through a central intermediary such as Visa or CB. First comes authorization, then clearing and settlement later in the day through the parties involved, up to the central bank, to make the payment final. It’s an older scheme. Authorization is quick, but final settlement can take several days.
:::Which payment methods are most vulnerable?
:: Jonathan :: That brings us to the cyberattack question. If I’ve understood correctly, the rails work differently. Which is most vulnerable? Is there one particular weak point, or can any rail be attacked?
:: Michel :: Having several rails is what makes payments resilient. If one track is blocked, the others can still carry traffic. There are also flows most people never see, particularly high-value payments that use an entirely separate rail operated by central banks.
:: Jonathan :: Which is most vulnerable, though?
:: Michel :: The latest figures from the Banque de France’s Observatory for the Security of Payment Means give us statistics on fraud. Fortunately, the observatory is there and keeps a close watch on all the banks in France.
Cards see the most fraud, unfortunately. That’s mainly because they’re less secure online. At the bakery, a physical card with a chip works very well. In-person payments have the lowest fraud levels. That’s what cards were designed to handle securely. Online, they work much less well, which is why we have measures like 3-D Secure, one-time passwords, and standards like PCI DSS. Cards alone account for 90% of fraudulent transactions in France.
:: Jonathan :: That’s enormous.
:: Michel :: By count, sorry. I mean the number of fraudulent transactions. These are small transactions.
:: Jonathan :: Of course. What about larger sums? Is there a risk there too?
:::Fraud and instant transfers
:: Michel :: I should qualify what I just said. The Banque de France has recently noted a rise in fraud involving instant payments and bank transfers. Why? We’ve made them much faster. With a card payment, you can dispute the transaction. It’s a well-developed system with security and insurance arrangements. You can call the bank, dispute the payment, and it may refund you. There’s no comparable cancellation mechanism for an instant payment. If I send someone €100,000 by mistake, the transfer goes through immediately and is irrevocable. I can’t just cancel it and get the money back.
Fraud involving instant payments is growing fast, particularly through impersonation and social engineering. Someone might call you and say, “Pay me using this QR code and I’ll invest the money for you,” or something along those lines.
:: Jonathan :: Can we really call that a cyberattack, though? It’s clearly social engineering.
:: Michel :: It uses digital tools and social engineering. The two go together. Typically there’s always some social engineering or intelligence-gathering behind an attack. You have to understand what’s in the systems involved and look at the risks we face.
:: Jonathan :: You’ve anticipated my next question. Could you list those risks and explain what banks do to protect themselves?
:::How banks defend themselves
:: Michel :: There’s separation between activities and between bank accounts. The core banking system sits inside what amounts to a fortress. Online and mobile banking are kept on separate paths, and money moves in and out over different payment rails.
Attacks have become so sophisticated that even all those defenses can fail. Vulnerabilities are discovered faster than banks can always keep up with patches, and modernization is difficult. Setting up a SIEM system or real-time monitoring is expensive. Sometimes monitoring lags by days. Teams audit logs and look at what happened in the past few hours, but what we need is the ability to act in real time. If €100 million is going out the door, you need a button that stops it immediately. In practice the chain of decision-making is often very, very long. There are other avenues of attack too.
:: Jonathan :: Given that risk, how do banks protect themselves in practice? I imagine there are governance arrangements to contain it. Can you be more specific?
:::Fraud protection and insurance
:: Michel :: There are different layers of protection. On the prudential side, banks keep reserves and have insurance arrangements. Fraud departments use those reserves to cover fraud. Say Jonathan calls because his card was stolen, a fraudulent payment went through, and he wants to stop it right away. Sometimes the money has already left the account and the bank can’t stop it. It wasn’t your fault, so the bank has to reimburse you from the reserves it set aside. I attribute that requirement to the Banque de France.
A small portion of card payment fees also goes toward insurance. If your flight is canceled, you may be able to call your bank even if you didn’t know you had coverage. You tell them you paid €2,000 for the flight, and the bank might say, “You’re covered for €1,500.” In that case you may get the money back almost immediately.
:: Jonathan :: So without realizing it, part of what we pay goes into that insurance pot.
:: Michel :: Yes. I’m known for saying my industry is the richest in the world because it charges too many fees. But some of those fees do pay for fraud coverage and reimbursements.
:::The four-eyes principle and a bank’s duty of care
:: Michel :: There are other safeguards, especially in governance. Transactions can require sign-off from different people under what’s sometimes called the four-eyes principle. One person can approve it, but they can’t do it alone. Another person in a separate role and with a lower level of risk must also sign off, depending on the thresholds. Those thresholds are gradually going up as everything speeds up. We want instant payments, so we raise the ceiling. You can now make an instant payment of more than €100,000, whereas a transfer could previously take two days. That gave us two days to try to get the money back.
There’s also the bank’s duty to advise customers. Sometimes the bank blocks a transaction and customers complain. They think it’s their money and none of the bank’s business. But the bank has a legal duty to advise you. It might ask, “Jonathan, you’re buying a lot of crypto. Are you sure?” If you say yes, it needs to check that you understand what you’re doing and that it wasn’t a mistake made by your child using your banking tools. So there are times when it’s legitimate for a bank to look at what its customers are doing. There are occasional oversteps, but these measures are intended to protect people.
:::What happens in a worst-case scenario?
:: Jonathan :: I’d like you to imagine a worst-case scenario. I don’t know whether you can answer this, or whether it’s covered in bank protocols. Say a massive attack knocks out payments, regardless of where it comes from. No card payments, no transfers, no way to pay. What happens? Have banks prepared for that kind of catastrophe?
:: Michel :: The first thing we build in banks and central banks is resilience. Before getting into worst-case scenarios, I want to reassure people about that. In my view, no sector is stronger on cybersecurity than banking. It spends more on it than any other industry, far more than healthcare, energy, or agriculture.
By resilience, I mean an entire region or even a country could go down and Europe could keep operating. Even if France and its central banking system went down, or even if the ECB were offline, the euro could keep working across Europe. I can only think of one incident of this kind. TARGET, which provides final settlement for payments in Europe, went down for a few hours. I don’t remember the exact date.
:: Jonathan :: That was quite a blow. We’ll put that story on screen in the video. It was remarkable.
:: Michel :: I mentioned that, as it happened, Bitcoin processed the largest transaction in its history around the same time. I don’t remember the value in billions of euros. The transaction went through in ten minutes at almost no cost. So yes, there have been major incidents, but resilience has allowed us to get back on our feet. In our industry we talk about a recovery point objective, or RPO, of zero data loss and a recovery time objective, or RTO, of under an hour. The aim is to be up and serving customers again within an hour. We can criticize banks, but they spend a great deal on resilience.
Let’s still consider a worst-case scenario, because it is possible. The Titanic sank. We don’t want the banking industry to go down, but in a disaster where we couldn’t recover for a week, we’d have to deal with the losses, fix the problem, and repair the damage. That’s why, as I recall it, Christine Lagarde urged Europeans to keep several days’ worth of cash at home. Whatever criticism you have of the banks, they’re already thinking about that kind of scenario.
:: Jonathan :: It’s striking. You mentioned this earlier. Do we have any idea how much cash to put aside?
:: Michel :: It depends on the household and what it needs, but I recommend at least a month.
:: Jonathan :: Right.
:: Michel :: I really do recommend keeping a month’s expenses at home. You could lose it or have it stolen, of course. But an outage at your bank wouldn’t cut off your access to it. It’s cash, central bank money.
:::Access to money during an outage
:: Michel :: If your bank is ever down for any reason, you may lose access to its services. La Banque Postale had incidents around the end of 2025 and the beginning of 2026. It suffered a denial-of-service attack that brought down online banking. Fortunately, it was resolved after a few hours, or perhaps a few days. Had it lasted longer, its customers would have been unable to access their money.
There was another problem, in 2025 or 2024, if I remember rightly. Payment services were down at Carrefour and other large retailers. Nothing worked. We never found out whether it was an attack or an internal incident because no one published a full analysis. It’s conceivable that some attacks have been described simply as incidents.
The euro is legal tender, so merchants generally have a legal obligation to accept it.
:::What guarantees can and cannot cover
:: Michel :: Normally, guarantee funds are in place if people lose money at a bank. But the Arkéa example was a hack of the bank itself. As I understand it, the money that went out belonged to the bank. It wasn’t Michel’s or Jonathan’s account that was drained. It was a major account belonging to the bank. About €100 million, or roughly $130 million as I estimated it here, was siphoned off. We’ll come back to what may have happened.
I see it as historically significant because there’s never been a digital heist of this kind. It’s like criminals forcing open an electronic vault and draining it. Fortunately, the bank is solid. The amount is tiny relative to its funds. But there are risks, including potential penalties, and bringing its governance and risk management up to standard will be expensive.
Someone could impersonate Jonathan or compromise his phone and get past strong authentication. His account could then be drained by an instant transfer made in his name. He’d have to ask the bank to reimburse him. It may be able to do so, unless a very large number of accounts are drained. Banks normally have internal alarms to flag excessive outflows and stop them. In the extreme scenario you’re describing, those alarms would also be down, like in a Mission: Impossible film. If many accounts were drained and the bank couldn’t reimburse its customers, a guarantee fund linked to the central bank might step in to help. If a major bank were hit, the ECB could also become involved.
There was a guarantee fund in Lebanon. I don’t want to scare people, but Lebanese depositors thought their money was safe in the bank. If their bank had a problem, they expected the Banque du Liban to cover them in full up to about $100,000. In a major crisis, the guarantee fund can stop working too.
:: Jonathan :: If you’d like to learn more about wealth management and how Bitcoin fits into it, we’ve launched a course on the subject. See the video description for details.
:::Artificial intelligence and cyberattacks
:: Jonathan :: Let’s move on to the second part of the podcast. It’s another issue in the news that we discussed on a previous episode with Fabrice Epelboin. We talked about breaches of government data and the growing power and impact of artificial intelligence. It’s already impossible to ignore, and I think it will matter even more in the years ahead. What’s your take on this potential technological time bomb in the context of cyberattacks?
:: Michel :: We haven’t established that AI sped up all the attacks we’ve seen in Europe. Take the attacks on refrigeration systems in France and several other European countries. They happened again this week. Over the past three weeks they started in France and appeared again in Spain, Denmark, and elsewhere. They’re still going on. We don’t know if AI was behind them, but it may have made them easier to scale.
There were also the bomb threats on planes. I’ve just got back from Madrid and wondered whether there might be a bomb scare on my flight. That kind of risk still feels new in Europe. We weren’t used to it.
There was also the theft of highly private or sensitive data from France’s public finance directorate, the DGFiP, and other public institutions. And there were attacks involving Binance, Ledger, Arkéa, and many others.
:::The race to find and patch vulnerabilities
:: Michel :: AI has a dual role. It helps find vulnerabilities faster.
:: Jonathan :: So to some extent it helps the defender.
:: Michel :: It helps the attacker too. If an attacker finds a flaw, they may exploit it before the defender finds and fixes it. The race between them is getting tighter. AI helps attackers discover new vulnerabilities and exploit the ones they find more quickly and at greater scale.
I work in cybersecurity, and I’ll remind people that, for a cybercriminal, mounting an attack takes work. It can take many person-days and sometimes a whole team. Take a bank that uses cloud containers to isolate parts of its systems and limit risk. The isolation layer can have vulnerabilities of its own. Exploiting one takes considerable expertise. An attacker may be trying to get inside, run code, stay undetected, and keep the system running. Their objective may be to steal something. They’re after more than a denial of service. That can take a lot of work, sometimes up to six months. In the meantime, defenders can find the vulnerability, develop and publish a fix, and give companies enough time to apply it. The attack then never happens.
AI can compress that timeline substantially. It becomes faster to find a vulnerability and launch an attack.
At central banks, for instance, we monitor components and get alerts when a vulnerability is published. When a patch is available, we have service-level agreements and deadlines to deploy it quickly without interrupting the system. Otherwise, we need a workaround. We can isolate or replace the component while we wait for a fix.
Banks need to adopt AI much faster, despite serious concerns about data sovereignty. Which AI system can we use to keep sensitive data from French banks and central banks inside Europe?
:: Jonathan :: That’s true.
:::Deepfakes and social engineering
:: Jonathan :: The other side of AI that’s all over social media is deepfakes. Looking at what they can do now, it’s almost impossible for a person to tell the difference. That brings us to social engineering, which AI can make easier. Someone could create a deepfake of my mother asking me to transfer a few thousand euros to her account, or use any number of tricks to manipulate me.
:: Michel :: To persuade you in the heat of the moment.
:: Jonathan :: Right. When it feels urgent, voice impersonation can be a real risk, even over the phone. People need to understand how effective these attacks can be. What do you make of it, and how do we protect ourselves?
:: Michel :: The first thing is awareness. We need to talk to my mother, your mother, and each other. Even with everything we know, we sometimes fall for things. So how do we get better at spotting them?
There are experts who can tell when something has been generated by AI, and they may be able to avoid the trap. Other people will fall for it. It comes down to how alert you are and how well you’ve prepared the people around you.
:::What to do when a call feels suspicious
:: Michel :: We get bombarded with calls. Even if the voice is convincing and you think it may be a real person, you can say, “I’ll call you back later. What company are you with? Leave me your details and I’ll get back to you.” The point is to avoid doing anything on the spot. If the contact came out of the blue, decline the request and decide later whether to act.
You might say, “I’ll log in later. Send me the information.” The problem is that what they send may include links to sites that can infect or compromise your phone or computer. So ask where you can find the information yourself. If it’s on the company’s website, go there directly. Unfortunately, this makes things harder for legitimate callers too, but we do need to stay alert.
As I see it, your banker would make a proper appointment to discuss a service, or ask you to reply later by email if you’re busy. My own rule is to avoid being available on the spot. I’ll follow up later, on my own initiative.
The second thing is to contact the person through another channel, whether you don’t know them or you suspect they’re impersonating someone you do know.
At Valuechain, we use AI too. We have a pool of around 300 consultants whom we help work independently. We contact them ourselves or with a degree of automation, always with our approval. Quite often we reach out on one channel and people who work in cybersecurity follow up with me on LinkedIn, by email, or elsewhere to check that the message really came from us.
:: Jonathan :: That’s remarkable.
:: Michel :: Absolutely. I tell them it was a good thing to do, and I confirm the message was ours.
Of course, someone determined enough might control several channels and answer on all of them using AI. So stay alert, be more cautious about what you trust, and ask advice from people around you.
Consulting is a difficult profession. Forgive the term, but when I started out, I sometimes saw it as a “bullshit job.” It can also be valuable. You can see that value by talking to people you know who have expertise. There’s no need to hire a consultant for every question. When you ask questions and have a network you can rely on, you’re stronger than the network on the other side. Raise awareness and stay in touch with people you trust.
:: Jonathan :: Right.
:::AI and bank authentication
:: Jonathan :: Let’s make this concrete. When you log into your bank account today, you generally use a password and a second factor, what we call 2FA. That could be an app such as Google Authenticator, a code by email, or a text message. With enough computing power, could AI try billions of combinations and get past both the password and the second factor? And even once someone gets in, a transfer requires another authentication check.
:: Michel :: If you compromise the same mechanism once, in principle you could compromise it a second time. But to reassure you, my answer is no. Let me explain.
:: Jonathan :: Why not?
:: Michel :: The one-time code is random. It isn’t a number you can simply work out. The exception would be if you already knew the bank’s random-number generation system and the seed it used, for example because someone involved in building it had been compromised. The seed is the starting point from which the other numbers are generated. That’s an extremely unlikely scenario with a very high impact. If you got that far, you’d already be inside the bank. In the scenario I have in mind, you could go straight for the money, as I suggested in the Arkéa example.
Let’s stick with your example of a one-time code. An SMS code can be intercepted. That can happen without AI or brute-force computation.
:::Strong customer authentication under PSD2
:: Michel :: In Europe we have the second Payment Services Directive, PSD2, and a third version is on the way. PSD2 sets out rules for strong customer authentication.
One element is possession. Do you have the phone that you authorized for this kind of transaction? If you’ve bought a new phone, you may have to authorize it before the transaction will go through.
In the example I’m describing, a criminal obtains your SIM and receives your text message but lacks your registered phone. The way I explain it, the bank has to refuse the transaction. If it lets it through, the bank is responsible and you’d be reimbursed in full, whatever the amount.
With an app-based second factor, the bank uses an app similar to Google Authenticator or Microsoft Authenticator to display temporary codes for these transactions. I describe this as a move away from SMS under PSD2.
:::Online banking, identity theft, and transfers
:: Michel :: There have been breaches of online banking. Online banking is separate from core banking, the system at the heart of the bank where the money is, in a manner of speaking, kept. Getting into online banking doesn’t mean getting into the core banking system.
:: Jonathan :: What can someone do at that stage?
:: Michel :: They would still need to impersonate you to make a transaction go through. Assume they already have access to online banking.
:: Jonathan :: When you talk about identity theft, you mean they’ve obtained personal details such as your first name, surname, and date of birth. You see that kind of scam a lot on Leboncoin. Put up a listing, say for a property, and someone asks you to send personal documents.
:: Michel :: Yes.
:: Jonathan :: So that theft of personal data lets them assume your identity.
:: Michel :: You’re getting me to describe how to attack a bank.
:: Jonathan :: Yes, maybe that’s a bad idea.
:: Michel :: I can give you a few points about defending against this kind of attack.
:::Payee checks and transaction limits
:: Michel :: Suppose someone tries to transfer money to a bank account you don’t recognize. That’s the sort of transfer the bank will flag. In the scenario we’re discussing, assume the attacker already has access to your phone and online banking and has managed to obtain the credentials needed to approve the transaction.
Normally you should receive an alert on your phone. It tells you a transaction is in progress and gives you a chance to stop it if you didn’t initiate it. If the phone is lost or stolen, someone else may be able to approve it. But if they try to send money to a new IBAN, the bank can stop it because the payee is missing from your list. You have to add the payee first. It’s an extra step and takes time, but it helps prevent large-scale fraud. If the same suspicious account has already been added elsewhere, it can be flagged and blocked. Banks already have strong protections against this kind of fraud.
Suppose someone has successfully impersonated you and approved the payee. They then try to send €100,000 by instant transfer. That’s theoretically possible, but your bank adviser may have set a much lower limit for your account. A request for a higher limit or a large transfer to an unfamiliar account may need the adviser’s approval. Stolen personal data can undermine identity checks, so there’s a human decision about whether to authorize the transfer or the higher limit. Responsibility can be shared. You can dispute the transaction, and the bank will often lose the dispute if it was the one that authorized it. There are more complicated scenarios, but the system today is well protected against this kind of attack.
I doubt the Arkéa incident followed those steps. My guess is that, as I suspect in the case of Binance, it involved a backdoor and someone with highly privileged access.
:: Jonathan :: OK.
:::Could Bitcoin help in a banking crisis?
:: Jonathan :: Let’s turn to the last part of the podcast and the main subject of this channel, Bitcoin. Back to the worst-case scenario. The banking system has collapsed, or at least it’s temporarily unusable. Bitcoin might offer a practical alternative because it operates outside that system. That’s at the heart of its promise as a peer-to-peer cash system. What’s your take?
:: Michel :: We’ve talked about this before. Bitcoin shouldn’t have to wait for a crisis to say, “Hey, I exist. Love me or adopt me!” A crisis wouldn’t be the holy grail, either. Adoption in that setting wouldn’t mean Bitcoin automatically takes over the world. It has value without a crisis.
In the scenario we’re discussing, it depends on how events unfold. Yes, Bitcoin can absolutely be useful, but you have to prepare. If you keep your bitcoin on an exchange that gets hacked or goes offline, you may be unable to use it. We’re imagining that the attack has affected banks and some exchanges as well.
:: Jonathan :: Sorry to interrupt, but that’s where self-custody matters. With actual bitcoin under your control, you can use its properties. With something like a Bitcoin ETF, you’re getting exposure to its price.
:: Michel :: I want to talk about both cases, holding through an ETF and holding in self-custody.
:::What a Lebanese currency exchange taught me
:: Michel :: First, an anecdote. I use examples from Lebanon because it’s my home country and I’ve just come back from a holiday there. I saw something I hadn’t seen for a long time, a place that exchanges Lebanese pounds, dollars, and euros. One thing that strikes me as odd about crypto exchanges is that they hold your money. You wouldn’t ordinarily leave your money with a currency exchange.
In Lebanon, towns like Jdita and Chtoura are historically known for their currency exchanges. Back in the 1930s or 1950s, Americans used to say children were carrying $10,000 to $100,000 in the street because the country was so safe. It was face to face, peer to peer. You turned up with, say, $10,000 or $100,000. You handed it over and the child converted it into Lebanese pounds, doing the arithmetic without a calculator and without making a single mistake. I saw that kind of exchange again myself. It still exists.
I went to exchange a few euros for dollars and Lebanese pounds. I handed the money to the man in front of me and expected questions about KYC, KYT, KYB, or the source of the funds. After all, it was an official business registered with the Banque du Liban. He checked the notes were genuine, counted out the money, put it in front of me, and said goodbye.
I thought, “That’s too fast. Surely it ought to take longer.” I’m used to things taking longer in France. This was faster than a one-click purchase. You hand over the money, he counts out the amount without making a mistake, and you leave. It was almost exactly the exchange rate I’d seen online.
Why bring this up? Because the business doesn’t act as a custodian. It makes a commission on the exchange. It doesn’t need anything else to earn a living, and it has no interest in keeping your funds. It isn’t a bank. So from my point of view, it’s unusual that Coinbase, Kraken, or Binance act as custodians. People ask for that service because self-custody in Bitcoin can be complicated, but to me self-custody is the natural arrangement.
:::Self-custody in a crisis
:: Michel :: You brought up self-custody. To me, it’s the most natural way to hold Bitcoin, and it can make all the difference in a crisis. It comes with risks, of course, but here’s why it can help. If something goes wrong and your money is on the exchange, you may be unable to withdraw it, just as you might be unable to withdraw from a bank account. Worse, the exchange can be drained, and you may have no insurance to cover you. In banking, you might expect support from the bank or central bank. There’s no equivalent backstop for Bitcoin.
So the natural thing to do with Bitcoin, in my view, is to keep control of it yourself. That carries significant risks too. There are data leaks, and there are reporting requirements being discussed that haven’t been voted on yet.
:: Jonathan :: That’s a separate issue, in a way.
:: Michel :: But it’s still the natural thing to do, and despite the risks I think it’s more worthwhile than leaving the money on an exchange.
How would it work? I suggest having a Bitcoin wallet and a Lightning wallet so you can transact. If you wait for the crisis to start, the person on the other side may have neither bitcoin nor a wallet and may be unprepared to use either. That’s why I say Bitcoin has to win that battle before a major crisis. Some people say a crisis will give Bitcoin its chance to prove itself. I think it has to prove itself beforehand if it’s going to be widely useful when the time comes.
:: Jonathan :: That’s one reason we created Surfin’ Bitcoin, to help people understand this. At StackinSat we’ve said from day one that Bitcoin’s properties make it useful for saving and as a long-term store of value. Its original promise was also peer-to-peer electronic cash. Bitcoin’s Lightning Network makes it possible to make payments almost instantly.
:::Stablecoins and everyday payments in Lebanon
:: Jonathan :: I want to go back to Lebanon, because you brought up something interesting. We had someone you know well on the podcast, another guest from Lebanon [name unclear in the transcript]. We send him our regards. On an earlier episode we talked about Bitcoin, and he also mentioned the rise of stablecoins.
As I was saying, people tend to be reluctant to spend their bitcoin because they expect it to appreciate over time. They’d rather spend fiat, meaning euros, dollars, or other conventional currencies. That’s where people see a use for stablecoins and where we’ve seen transaction numbers grow considerably. I don’t have the figures, and I don’t know if you do, but I imagine they’ve caught on in Lebanon. Maybe it wasn’t an explosion, but they’ve become more common. In parts of Latin America, their use really has taken off. How do you see that trend and the role of stablecoins in day-to-day payments?
:: Michel :: In Lebanon, the attitude toward stablecoins is quite different. People are very attached to holding physical US dollars. I’ve seen that firsthand, and so has the guest you mentioned. The idea is to hold cash. The problem with cash, of course, is inflation and devaluation. Gold is another store of value for Lebanese people and perhaps for people in other countries too, such as India. And in a country like Germany, people still hold a lot of cash.
:: Jonathan :: Where do stablecoins fit into that?
:::The risks and limits of stablecoins
:: Michel :: Personally, I advise against them. That’s my view, and other people can have their own. It’s partly because I’m a bit of a Bitcoin maximalist, or perhaps more than a bit. But I also think stablecoins have a serious problem. I’ve made the case on this channel several times that they have an inflationary effect.
:: Jonathan :: I’ll point people to the video where Michel discusses those issues in detail.
:: Michel :: I won’t go through the whole argument again. Based on how adoption is expected to grow, my estimate is that stablecoins could add between 5% and 30% to inflation. In my view, prices could rise as stablecoin use expands. If you aren’t convinced, I’d point you back to that discussion.
The second point is that I see stablecoins as less well backed over the long term than some other assets. From that perspective, a cryptoasset linked to gold would be preferable.
Why? A stablecoin is stable relative to a currency that can itself lose value. You may be happy to hold stablecoins, but they can be frozen, and their transactions are traceable. In the Binance example, as I understand it, billions could be frozen because they were held in stablecoins. Stablecoins can be somewhat faster and more programmable than legacy payments. They’re another payment rail, one that runs on blockchains. I don’t see them as protection in a crisis.
:: Jonathan :: That brings us back to our worst-case scenario, a banking collapse. You might assume a stablecoin like USDT could be an alternative. I’m not a stablecoin specialist and can’t remember all the others.
:: Michel :: Some use Ethereum.
:: Jonathan :: They use ERC-20 tokens. So you might say they’re decentralized and separate from the banking system. But there’s a catch. Do we know why Tether, the issuer of USDT, chose not to comply with MiCA? I don’t remember exactly why.
:: Michel :: Maybe for the very reason you’re getting at. If Tether had a European branch regulated in Europe and subject to European requirements, it would also have to hold collateral there. That’s far from ideal for Tether, whose approach is to keep dollars in the US and buy US Treasury securities. It didn’t want to do that in Europe. It could simply be a business decision, or a bit of a standoff, like the ones between Apple and the European Commission. In this case, Tether versus MiCA.
:::ETFs and access to funds during a crisis
:: Michel :: Let’s come back to ETFs and stablecoins together. If you try to cash out stablecoins in a crisis, you may find them blocked just like a bank account. If the collateral is gone, you can still hold and transfer an ERC-20 token on Ethereum with your own key, but the token may be worthless. The tokens also run on smart contracts. In the cases I’m describing, regulated issuers may be required to freeze funds to keep liquidity under control while they look for a solution.
A lot of people ask me what they should do to diversify or build resilience. Should they move abroad or stay where they are? If they buy something, what should they buy? Some look at ETFs. I can see a case for ETFs that track the CAC 40, the S&P 500, or other indices. I see it differently for Bitcoin and other cryptocurrencies.
The thinking is that you still own a security if the bank fails. The underlying companies, such as Apple, Total, or Air Liquide, are still there, and you still own the security. But if you want to use that money, you have to sell the security and receive the proceeds in a bank account. That account may be blocked. Even if the market is doing very well and Total’s shares have soared, whether you sell for €10,000, €100,000, or €1 million, the proceeds may end up in an account you can’t use.
We’ve seen this in several countries. I don’t want to scare people, but there is genuine concern in Europe, including among bankers in France who see money going out. Some funds are coming in, of course, while other funds leave out of fear. My point is that, in that kind of crisis, securities won’t give you a way to pay. A Bitcoin ETF won’t either.
Self-custody gives you the freedom to pay directly. The opening of Satoshi Nakamoto’s white paper is about peer-to-peer electronic cash. In my view, Bitcoin was created for a crisis like the one we’re discussing, so people could pay one another directly without a trusted third party.
:: Jonathan :: Wonderful. Now…
:::Risks to Bitcoin and self-custody
:: Jonathan :: To put some of this in perspective, I’d like to spend the last few minutes on the risks to Bitcoin. We recently recorded another podcast on this, but let’s return to it briefly. AI and quantum computing could pose risks to Bitcoin too. Is the quantum risk still there, and could it grow over the next few years? How do you see it?
:: Michel :: Yes. We don’t have to compare every risk, but we can at least mention the supply-chain attack that, as I understand it, recently affected Ledger in Southeast Asia and involved other players too. We’re talking about self-custody, where you control your own bitcoin. Unfortunately, there’s a risk of kidnapping and unlawful detention in France and elsewhere. Someone can threaten you in an attempt to make you hand over your private key.
There are also cyber risks, including supply-chain attacks. A hardware wallet can be tampered with during manufacturing or distribution in a way that compromises the private key. As I describe it here, attacks like those have already occurred, with more than $100 million stolen. Diversifying into Bitcoin therefore comes with its own risks. That’s why I wouldn’t treat it as the only fallback, without getting into personal advice, which isn’t the aim here.
:::AI, quantum computing, and Bitcoin’s defenses
:: Michel :: There are other potential vulnerabilities. You mentioned AI and quantum computing. Could AI discover and exploit a flaw in Bitcoin? People have been testing that since the early days of AI and, so far, nothing has come of it.
Is the Bitcoin community aware of the problem? Yes. Can it respond quickly when it finds a vulnerability? It’s always a game of cat and mouse, and we use AI to defend ourselves too. I think the Bitcoin community is well ahead of banking in its use of AI for defense. Banks are looking at how to integrate AI, but in my view they haven’t got far enough to use it effectively for defense.
As an aside, banks and central banks are also trying to protect themselves against risks from AI itself. We want to deploy it in a way that preserves sovereignty. There’s a concern that a model trained outside Europe, even if we host it on our own infrastructure and control it locally, could make a mistake or behave in a deliberately harmful way and break something inside the bank.
Coming back to Bitcoin, I see AI as a relatively small risk. Bitcoin is relatively simple in many respects.
:::Recovery phrases and private keys
:: Jonathan :: I’d like to go back to my earlier example of someone trying to find a password. In Bitcoin, we have what’s called a private key. I was thinking of the 12 or 24 words, though I also think of a string of letters and numbers…
:: Michel :: The words aren’t the private key.
:: Jonathan :: Could AI, with all that computing power, make it more likely that someone finds the private key? I used to hear the odds were one in a million, or one in a billion. I don’t remember the exact figure. We’ll clarify the actual odds. Does more powerful AI increase the chances of finding it?
:: Michel :: The private key is extremely hard to find. I described the problem here as factoring a large number into two primes, with one involved in generating the private key and the other in generating the public key. [See technical note below.] What are these keys? The private key is a secret you keep to yourself. It lets you move funds by signing a Bitcoin transaction. Public keys are used to derive addresses. Loosely speaking, think of them as identifiers, or something like an account number or IBAN. You can generate many of them. This public-private key pair is what could be threatened, potentially by AI or quantum computing.
:::Cryptography and the quantum threat
:: Michel :: Banking, electronics, and the digital world, including the HTTPS connection you see on websites, depend on cryptography. If someone breaks these mechanisms, the consequences will extend well beyond Bitcoin. Some things are easier to break than Bitcoin and will probably be broken first. Bitcoin will see those failures coming, and that’s why everyone is preparing.
As a security professional, take TLS, the protocol behind HTTPS. We used to have SSL, then TLS 1.0 and TLS 1.2, and we’ve moved to TLS 1.3. As the standards evolve, they can use algorithms designed to resist a future quantum computer. The mathematics gives us tools for that. The question is who will be ready first, banks or Bitcoin. I’d almost bet that Bitcoin will get there first.
What could a quantum computer do? For instance, it could potentially take bitcoin from early outputs associated with Satoshi Nakamoto where the public key is already exposed. Some Bitcoin output types show a hash derived from the public key instead of revealing the key immediately. That conceals the public key. My point is that when the public key hasn’t been exposed, a quantum computer can’t immediately run the same calculation against it. [See technical note below.]
What do I mean by “break”? Recover the private key from the public key. There are already proposals for Bitcoin. One concerns the signature scheme, and another concerns how public keys are exposed in Taproot, including in the context of the Lightning Network.
It’s similar in banking. For example, Thales offers hardware security modules, or HSMs, in France to protect bank keys. I distinguish between being fully quantum-ready and already implementing certain key generation and management algorithms intended to resist quantum attacks. Banks, Bitcoin, Ethereum, and other actors are preparing. I still see the risk as very low. Based on the research I’m aware of, I think we have several years, perhaps five or ten. Personally, I think we may have longer. Others disagree. In my view, we should aim to be ready within five years so we have better protection when the time comes.
A quantum computer could theoretically break a signature and drain my wallet without my knowing. If the public key is still concealed, it doesn’t have that key to work from. Replacing the signature scheme with one designed to resist quantum attacks could address this risk, though blocks could get bigger. Engineers and the Bitcoin community are working on ways to do that without making the blockchain grow excessively or slowing the network. What I tell Bitcoin investors is that there’s no reason to panic today. Everyone is preparing, and I think Bitcoin can move faster than banks.
:::Quantum computing and mining
:: Michel :: Quantum computing could in theory deliver extraordinary computing power. Could that translate into 51% or even 60% of the effective mining power, and make attacks like double-spending possible? It’s complicated. The network is designed to adjust when hash rate rises, though it takes time to respond. People are already studying whether the mining rules would need to change to deal with a future quantum threat. They haven’t changed yet.
Existing mining ASICs are extremely powerful. The capacity of the entire Bitcoin network is such that, in my view, even a very powerful quantum computer would be small by comparison. So Bitcoin still has some breathing room.
:: Jonathan :: OK. It’s fascinating because you’ve opened up a whole other subject. Let us know in the comments if you’d like another episode with Michel about quantum computing and mining.
Thank you for accepting the invitation.
:: Michel :: Thank you.
:: Jonathan :: See you soon for another episode. Please like the video and share it with people you know. It matters a lot to us and to the YouTube algorithm. Thanks again, Michel. See you soon.
:: Michel :: Until next time.
:: Jonathan :: Goodbye, everyone.
---


Comments