In the Era of Claude Mythos, Where Is Financial Infrastructure Heading? Will Blockchain Become a 'Risk Diversification Destination' in Japan?

This is the continuation of the story.

The other day, NHK's "Close-up Gendai" featured a special on the cybersecurity threats posed by Anthropic's AI system "Claude Mythos".

The program introduced the danger that AI could discover unknown vulnerabilities at high speed, enabling more sophisticated cyberattacks than ever before. However, while watching the program, I had one question.

"In that case, which is more resilient in the AI era: centralized financial systems or blockchain?"

So, I had ChatGPT verify this question and deepened my own analysis.

As a result, several important points that were not discussed in the program became apparent.

What was discussed in the program was the threat to "centralized systems."

Current banks have

• ATMs

• internet banking

• core banking systems

• internal authentication systems

• data centers

, and many other systems that are closely linked.

AI like Claude Mythos has the potential to perform

• source code analysis

• network analysis

• OS analysis

• discovery of unknown vulnerabilities (zero-day)

at a speed that is overwhelmingly faster than humans.

If an attacker succeeds in gaining administrative privileges or infiltrating a core banking system, the impact could spread to the entire bank.

In other words, in a centralized system, the structure is such that a single critical entry point being breached can cause the entire organization to suffer a major impact.

So, is Bitcoin safe?

Here, I had the opposite question.

Bitcoin has tens of thousands of nodes around the world.

If there are tens of thousands of attack targets, isn't it actually more dangerous?

At first glance, that seems to be the case.

However, what I realized through discussions with ChatGPT is that

the 'number of attack targets' and the 'magnitude of the impact if an attack succeeds' are separate issues.

For example, in the case of a Japanese bank, if a single core banking system or authentication system is breached, it could potentially affect millions of users.

In other words, it is structured so that by breaching a single entry point, one can cause significant damage to the entire system.

On the other hand, the situation is different for Bitcoin.

The tens of thousands of nodes that exist around the world each independently verify the blockchain.

Even if an attacker were to completely hijack one node, that node would only distribute abnormal data.

Other nodes, because they verify blocks and transactions themselves, will not accept the fraudulent data.

In other words, by compromising just one node, one cannot rewrite the entire Bitcoin ledger or steal assets from users around the world.

Of course, having tens of thousands of nodes means there are that many more targets for attempted intrusions.

However, in exchange, the 'value of compromising' a single node is small.

Conversely, in a centralized system, even if there are relatively few entry points, a single breach can potentially cause the entire organization to cease functioning.

This difference is precisely why

The 'number of entry points' and the 'impact of breaching an entry point' are separate issues

is what that means.

However, the AI era changes the story a bit.

That said,

'Because it's Bitcoin, it's absolutely safe'

is not necessarily the case.

What if Claude Mythos were to

analyze the Bitcoin Core source code and

discover an unknown vulnerability shared by nodes around the world?

Defects that humans might not have noticed for years in the past,

AI might find in just a few hours.

In other words,

in the AI era,

vulnerabilities in common software

will become more important than ever.

Ethereum is even more difficult.

Ethereum's situation is different from Bitcoin's.

The reason is

smart contracts

.

On Ethereum,

• DeFi

• NFT

• DAO

and millions of other smart contracts are running.

These are all programs written by humans.

In other words, Claude Mythos has the potential to discover a massive number of

• reentrancy attacks

• authorization management flaws

• overflows

• logic errors

and more.

It is not the blockchain itself, but the

application layerthat will become the new target of attacks in the AI era.

JPYC is no exception.

Stablecoins like JPYC are also

implemented as smart contracts on Ethereum.

In other words,

it is not safe just because Ethereum is secure;

if there is an unknown flaw in JPYC's own contract,

There is a possibility that AI could exploit that.

In fact, the most dangerous element might be the exchanges themselves.

What was interesting here was that,

rather than Bitcoin itself,

crypto asset exchanges like Coincheck and bitFlyer,

are the ones whose risk increases in the AI era.

This is because,

exchanges are, after all,

just ordinary IT companies.

• Web servers

• internal networks

• databases

• authentication systems

As long as they are centralized systems that possess these,

they can become the very targets that Claude Mythos excels at attacking.

Even in past crypto asset theft incidents,

it was not the blockchain itself that was compromised,

but the peripheral systems that were breached.

The concept of hardware wallets

Conversely,

Hardware wallets like Ledger and Trezor

keep private keys offline.

Private keys disconnected from the network

are less susceptible to direct attacks than online systems.

Of course, they are not a panacea, but

the concept of 'keeping private keys disconnected from the internet'

is an idea that

will become increasingly important in the AI era.

The reason why cryptocurrency has become widespread in developing countries

Currently,

in many countries where Bitcoin, USDT, and the like are used as social infrastructure,

the banking system is immature.

• There are few ATMs

• People cannot hold bank accounts

• The local currency is unstable

• International remittances are difficult

Due to these circumstances,

blockchain is functioning asa substitute for banks.

However, the situation is different in Japan.

Japan possesses one of the world's leading financial infrastructures.

ATMs, bank accounts, and cashless payment systems are all well-established.

Therefore,

the scenario of 'using Bitcoin because there are no banks'

does not apply.

However,

in the AI era, the criteria for evaluation may change.

In the AI era, 'resilience' will become more important than 'efficiency'.

Until now,

the premise was that banks are convenient and safe.

However,

in an era where AIs like Claude Mythos possess advanced cyberattack capabilities,

the danger of relying on a single system

may be reconsidered.

In other words,

rather than efficiency,

resilience (the ability to recover and adapt)

will become important.

The same concept as disaster preparedness

This is easy to understand if you compare it to electricity.

Normally, we use commercial power.

However,

In the event of a disaster,

• private power generation

• solar power

• storage batteries

are combined.

Might finance be the same?

Normally using banksand,

if a large-scale cyberattack shuts down the banking system,

a separate financial infrastructure called blockchainwould function.

This kind of

idea of using multiple financial infrastructures in parallel

might become important.

The possibility of becoming a risk diversification destination in Japan as well

I believe that,

in the AI era,

cryptocurrency will not be

a 'substitute for banks'in Japan,

but rather,

a 'risk diversification destination for financial infrastructure'

I believe there is a possibility it will be re-evaluated as such.

For example,

• receiving salary via bank

• daily life also via bank

• some assets in Bitcoin

• stablecoins for yen-denominated payments

• using multiple payment methods in emergencies

This is the form it would take.

This is not a rejection of banks.

Rather,

it is a way of thinking that does not rely on a single system.

Finally,

in the NHK program,

the threat of cyberattacks brought about by Claude Mythos was introduced.

However,

beyond that,

the design philosophy of financial infrastructure itself might change

I felt that the discussion did not go as far as that point.

Until now, blockchain has

mainly spread in countries with weak financial infrastructure.

However, in an era where AI dramatically increases the ability to attack centralized systems, the situation may change.

What is important is that

"Banks or blockchain"

is not a binary choice.

In the AI era,

the efficiency of centralized systemsand

the fault tolerance of decentralized networksmust be combined.

What will truly hold value in the future of finance is not the "most convenient system," but perhapsa mechanism that allows society to keep running even if one part fails.

Claude Mythos has not merely raised the issue of cybersecurity.

It seems to be posing a more fundamental question to us: "What is financial infrastructure?"