Kraken Security Labs just 15 minutes to hack both of trezor's crypto hardware

[bL4nkcode]

Kraken Security Labs has devised a way to extract seeds from both cryptocurrency hardware wallets offered from industry leader Trezor, the Trezor One and Trezor Model T.

The attack requires just 15 minutes of physical access to the device. This is the first time that the detailed steps for a current attack against these devices has been disclosed.

Twitter Post: https://twitter.com/krakenfx/status/1223253508956266496
Source: https://blog.kraken.com/post/3662/kraken-identifies-critical-flaw-in-trezor-hardware-wallets
Trezor response to the attack: https://blog.trezor.io/our-response-to-the-read-protection-downgrade-attack-28d23f8949c6


Btw I'm a ledger user and never had trezor HW yet.

Note: Don't ever let anyone have touch/physical access any of your hardware wallets. Keep it safe always.

[Zicadis]

Well these kind of attacks were always going to be possible. For example, anybody who has access to your hardware wallet could simply install a device that records your PIN and transmits it wirelessly, or install cameras in the room or pressure-sensitive film over the buttons to register your presses and record the PIN.

They're still going to be a step up over simply storing your private key or seed phrase on a paper wallet.

[The Sceptical Chymist]

Wait, what?  From Trezor's response, this has been known about since October of last year (although they apparently just responded now).

Anyway, I'm ignorant as far as technical details go but this stood out to me:

It’s important to note that this attack is viable only if the Passphrase feature does not protect the device. A strong passphrase fully mitigates the possibilities of a successful attack.

So that tells me that it isn't always possible to hack the Trezor if someone has possession of it.  Someone please correct me if I've interpreted that incorrectly.

This is still very interesting to me, as I've often wondered how easy it would be for someone to get access to a hardware device's private keys.  Till now I had no clue, and I'd assumed that it would have been impossible--oops!  I've never used a Trezor, but now I'm curious as to what it would take to hack a Ledger or any of the other hardware wallets on the market.

[mgoz]

So that tells me that it isn't always possible to hack the Trezor if someone has possession of it.  Someone please correct me if I've interpreted that incorrectly.

You are correct. This attack is not possible if you use a passphrase.

From Satoshi Labs:

To protect our devices against physical attacks without compromising our principles, we invented and implemented the Passphrase feature. The passphrase itself is not stored anywhere in hardware, SatoshiLabs doesn’t possess a backup, and therefore cannot be exposed or in any way “hacked” by a third party. When it comes to the passphrase, the user is the most crucial part of the whole process as it’s up to you to decide how complex your passphrase will be, how will you store it, protect it, or whether you should use one at all.

From Kraken:

Enable Your BIP39 Passphrase with the Trezor Client
This passphrase is a bit clunky to use in practice but is not stored on the device and therefore is a protection that prevents this attack.

[Zicadis]

Wait, what?  From Trezor's response, this has been known about since October of last year (although they apparently just responded now).

Anyway, I'm ignorant as far as technical details go but this stood out to me:

It’s important to note that this attack is viable only if the Passphrase feature does not protect the device. A strong passphrase fully mitigates the possibilities of a successful attack.

So that tells me that it isn't always possible to hack the Trezor if someone has possession of it.  Someone please correct me if I've interpreted that incorrectly.

This is still very interesting to me, as I've often wondered how easy it would be for someone to get access to a hardware device's private keys.  Till now I had no clue, and I'd assumed that it would have been impossible--oops!  I've never used a Trezor, but now I'm curious as to what it would take to hack a Ledger or any of the other hardware wallets on the market.

Isn't the passphrase just an additional word added to the seed phrase? If so, surely they could just retrieve the 24 word seed phrase, and then brute force the 25th word via a dictionary attack?

Doesn't really seem to be that much additional security, or is there something I'm missing?

[mgoz]

Isn't the passphrase just an additional word added to the seed phrase?

It is used in combination with the recovery seed to access a single hidden wallet and can be any length you want. You can also create as many hidden wallets as you want with multiple different passphrases. Technically speaking, yes you could attempt to brute force a passphrase for a single hidden wallet with the recovery seed taken from the chip, but this would be extremely difficult unless it is a very basic passphrase. Any incorrect passphrase is still going to generate a wallet, but with no funds. Using a passphrase that could be easily brute forced would defeat the purpose of using a passphrase to begin with unless you are creating decoy wallets. A random 12 character alphanumeric passphrase would cost $128 billion on average to crack today. Passphrases will weaken over time, so choose strong ones, or transfer funds to another hidden wallet with stronger passphrase if existing is too weak.

[Pmalek]

If so, surely they could just retrieve the 24 word seed phrase, and then brute force the 25th word via a dictionary attack?

Brute forcing a long and complicated password would take years. Your password should never consist of easy to guess dictionary words. I am not sure how reliable this source is https://www.betterbuys.com/estimating-password-cracking-times/.

But they say that a 10 character password would take 4 months to brute force.
An 11 character password 10 years, and a 12 character one would take 200 years.

I like the part where you can test the strength of passwords on site. Just don't enter passwords you use in real life. But an interesting feature to play with. 

[Saint-loup]

Kraken Security Labs has devised a way to extract seeds from both cryptocurrency hardware wallets offered from industry leader Trezor, the Trezor One and Trezor Model T.

The attack requires just 15 minutes of physical access to the device. This is the first time that the detailed steps for a current attack against these devices has been disclosed.

Twitter Post: https://twitter.com/krakenfx/status/1223253508956266496
Source: https://blog.kraken.com/post/3662/kraken-identifies-critical-flaw-in-trezor-hardware-wallets
Trezor response to the attack: https://blog.trezor.io/our-response-to-the-read-protection-downgrade-attack-28d23f8949c6


Btw I'm a ledger user and never had trezor HW yet.

Note: Don't ever let anyone have touch/physical access any of your hardware wallets. Keep it safe always.

Where have you seen that?   Did you read the report of the exploit and watch the video? https://blog.kraken.com/post/3662/kraken-identifies-critical-flaw-in-trezor-hardware-wallets
You need to fully disassemble the ledger and extract the chipset, the case sealing has to be broken.
In practice only an electronic engineer in a laboratory with the necessary equipment is able to conduct this attack. So I wouldn't be surprised if the leak came from a competitor in fact The exploit is highly technical and sophisticated.

[Lucius]

In his official press release, Trezor said this is very likely a similar vulnerability Ledger Donjon Team discovered last year and I post about this in this topic Trezor&Keepkey - Unfixable Seed Extraction - A practical and reliable attack!.

As you can see, there is no fix for this problem in such a way that it can be repaired with the new firmware. The only sure solution is to use a passphrase - " Donjon Team suggest that this passphrase should be about 37 characters long to prevent dictionary and brute-force attacks."

Given that passphrase is not stored in a hardware wallet, it will protect you even in the event of a physical attack when the attacker extracts seed from a hardware wallet.

[o_e_l_e_o]

From Trezor's response, this has been known about since October of last year (although they apparently just responded now).

The Ledger team released details of probably the same attack in July of last year. Certainly the attack they performed has the same outcome - if someone has physical access to your device they can extract the seed. See Lucius' link above. I own a Trezor device, but have since reset it, don't store any coins on it, and haven't used it at all, since a few weeks after that release. I was concerned about the attack, and I was also concerned about Trezor's response, which was essentially "Meh, use a passphrase", and didn't state in any way how they were going to address the issue (and they still haven't).

So that tells me that it isn't always possible to hack the Trezor if someone has possession of it.  Someone please correct me if I've interpreted that incorrectly.

So the attack is always possible regardless of whether or not you use a passphrase. An attacker using this method will always be able to extract your 24 word seed. The different is whether or not your coins are also secured by a passphrase. If they aren't, then once an attacker has your seed they can steal your coins. If they are, then once an attacker has your seed they can try to brute force your passphrase, and then steal your coins.

In practice only an electronic engineer in a laboratory with the necessary equipment is able to conduct this attack.

Quite the opposite. The Ledger release above shows the attack was possible using a single board with components costing less than $100, using only "basic electronics techniques". Not only is this attack potentially easily reproducible, but someone could also manufacture and sell those boards.

For a passphrase to be as secure as a 24 word seed phrase, which is the security level you have to reach if you want your wallet to be as secure as if this attack didn't exist, then it needs to be 37 random characters. Given that only a minority of users even use a passphrase, and of those who do, a very small minority of them will use a passphrase of 37 random characters, Trezor's response to this attack is wholly unsatisfactory. Since most passphrases in use are probably human generated, then like passwords, they will be short, not random, and bruteforcible.

[DaveF]

This was discussed before here:
https://bitcointalk.org/index.php?topic=5146284.0

Dave's view: Neither of the big 2 hardware wallet makers are that great against a determined foe. There are ways of mitigating the risk, but they are human dependent (long passphrase, keep it secure, etc.)

I am partial to the coldcard, but even then I know it is not perfect.

But, as stated before and by other posters in this thread 37 character passphrases are a joke.

100% personal opinion and nothing more: I trust NOTHING that comes from  SatoshiLabs / Slush and never will. I used to when I was "young and learning about bitcoin and crypto" but now that I know better. Nope. He / that team do just enough to get the job done and no problems are ever their fault.

-Dave

[hugeblack]

I don’t know what is the benefit of republishing this news, everyone knows that physical access to any device will expose you to danger and that the use of easy-to-guess passwords makes you vulnerable to hacking. Perhaps the platform is trying to gain media.

Everyone should remember that there is no wallet completely safe and there is nothing against hacking, your use of more than one wallet, strong passwords, distrust of any third party, follow-up news will spare you all of these hacks.

[Saint-loup]

In practice only an electronic engineer in a laboratory with the necessary equipment is able to conduct this attack.

Quite the opposite. The Ledger release above shows the attack was possible using a single board with components costing less than $100, using only "basic electronics techniques". Not only is this attack potentially easily reproducible, but someone could also manufacture and sell those boards.

So because Ledger, their main competitor is saying that (without any proof) it's necessarily true?
You have read the Kraken report and watched their video? You really think it's easily reproducible by anyone? There are certainly more chances to destroy the chip by doing all those manipulations than to succeed for the average handyman IMO.
And the equipment they used doesn't cost only 100$...

For a passphrase to be as secure as a 24 word seed phrase, which is the security level you have to reach if you want your wallet to be as secure as if this attack didn't exist, then it needs to be 37 random characters. Given that only a minority of users even use a passphrase, and of those who do, a very small minority of them will use a passphrase of 37 random characters, Trezor's response to this attack is wholly unsatisfactory. Since most passphrases in use are probably human generated, then like passwords, they will be short, not random, and bruteforcible.

Again you are quoting the Ledger claims... but you said exactly the opposite 3 days ago for just 2048⁴ combinations...  

For every possible seed phrase he has to go through 2048 iterations of PBKDF2 using HMAC-SHA512 to get a 512-bit seed number. For each seed number he then has to perform HMAC-SHA512 to create a master private key, again to create a child private key, ECDSA with secp256k1 to create a public key, SHA256 and RIPEMD160 to get an address, and then each address has to be checked for balance on the blockchain. And that's the minimum amount of work assuming you are using m/44'/0'/0'/0/0. Any other address, change address, account, or purpose (address type) requires additional work.

The passphrase is UTF8 encoded, 37 random characters gives 1 million³⁷ combinations...

[o_e_l_e_o]

So because Ledger, their main competitor is saying that (without any proof) it's necessarily true?

Ledger disclosed the attack to Trezor. Trezor's response was essentially "You need physical access and mitigated by a passphrase". If Ledger had lied and completely made it up, would Trezor not just have said that? It would be pretty irresponsible of Ledger to release the actual set up they used.

but you said exactly the opposite 3 days ago for just 2048⁴ combinations...  

All I said in that previous thread was that there was more work to be done than simply trying different seed phrases which would slow down the bruteforce attempt, and 10 billion combinations a second is too high a number. I never said a bruteforce attempt would be impossible. Assuming an attacker also knows your addresses (because otherwise why would they be targeting you), then with this attack you can also skip the blockchain look-up step.

2048^4 is 44 bits of entropy. This is brute-forcible. The average human generated password has entropy of around 40.5 bits. Again, this is brute-forcible. My point above is that if you are assuming your seed can be accessed, but you want to maintain the same level of security, then you need a very long and random passphrase, which almost nobody will be using.

[DaveF]

So because Ledger, their main competitor is saying that (without any proof) it's necessarily true?

Ledger disclosed the attack to Trezor. Trezor's response was essentially "You need physical access and mitigated by a passphrase". If Ledger had lied and completely made it up, would Trezor not just have said that? It would be pretty irresponsible of Ledger to release the actual set up they used.

Other people did, It's out there if you look.
Now this next little bit REALLY makes it look like I shilling for Coinkite / Coldcard but a blurb from their marketing:

SECURE ELEMENT FOR KEY STORAGE
We find it a quite scary that some Bitcoin wallets trust the main microprocessor with their most valuable secrets. Instead, Coldcard uses a Secure Element to protect your Bitcoin.

Specifically, the Coldcard (Mk3) uses Microchip's ATECC608A to store the critical master secret: the 24-word seed phrase for your BIP32/BIP39 wallet.

This little chip is very powerful. Communication is controlled by complex challenges and SHA-256 responses which prevent replay and eavesdropping. The secure element enforces cryptographically, that the attacker must know the PIN to access the secrets. An attacker cannot brute-force combinations or replay a previous login sequence. This remains true even if they removed the chip from the board or fully-replaced the firmware in the main microprocessor. In fact, even with the secure element removed from the system, and all the secrets of the main micro fully-known, the attacher would still only get 13 tries before the secure element bricks itself! (Don't worry, this counter is reset every time you login correctly.)

So if they can do it why can't others?

They also have the self destruct pin option and the sign with SD card so it's always airgaped option.

I would say it's not that difficult, but I am sure it is. But if they can do it other people can too. That is if they actually care enough to try.

On another note they also have a nice document about PIN use and some thoughts about how / why it works the way it does:

https://raw.githubusercontent.com/Coldcard/firmware/master/docs/pin-entry.md

-Dave

[Wind_FURY]

I don’t know what is the benefit of republishing this news, everyone knows that physical access to any device will expose you to danger and that the use of easy-to-guess passwords makes you vulnerable to hacking. Perhaps the platform is trying to gain media.

It wasn't republished. Kraken did their own version of an attack done in hacking conferences, and thought that it might be a good, which it is, to remind everyone, again, to put a passphrase.

Everyone should remember that there is no wallet completely safe and there is nothing against hacking, your use of more than one wallet, strong passwords, distrust of any third party, follow-up news will spare you all of these hacks.

Put a passphrase.

[HCP]

Yeah I saw this article on my newsfeed the other day... and was like "Wasn't this already done and discussed?"

Seems that Kraken just reproduced what Ledger already did... and said what we already know:
If you use a Trezor, make sure you are using the passphrase feature!


Although, the "requiring physical access to the device" part makes this "attack" somewhat theoretical for most people...

[DaveF]

Although, the "requiring physical access to the device" part makes this "attack" somewhat theoretical for most people...

Yes and no.
Where is your hardware wallet? Is it always attached to you? Or, is at home while you are at work?
Do you bring it with you on vacation or is it in a safe at home? etc.

It's a 15 minute attack, so if they know you have it they can probably get in and get it and get your seed (if your don't have a strong password) in less time then it takes to get dinner. Depending on who you ask a 8 character password is minutes.

-Dave

[HCP]

Assuming they lug some laboratory equipment with them while breaking into my secure apartment building... and they know precisely where my hardware wallet is stored... assuming, as you say, that they even know I have one in the first place!

That's why I said "somewhat theoretical"... Yes, it's possible... but is it really probable? For me... the answer would be "No". I think it would be more likely that my wallet/phone would be stolen from my pocket/bag while I was out and about during the day... and I'm not terribly worried about that either.

While I will admit that it is a huge flaw in the design of the Trezor, this is not exactly a doomsday scenario that everyone seems to want to make it... Put on a "decent" passphrase (which isn't that difficult or annoying to use) and the entire thing essentially becomes a moot point anyway.

For anyone with enough coin stored on a Trezor to be worried about losing it... then simply switching to a Ledger or Coldcard or another hardware wallet shouldn't be a massive issue.

[Wind_FURY]

Although, the "requiring physical access to the device" part makes this "attack" somewhat theoretical for most people...

Yes and no.
Where is your hardware wallet? Is it always attached to you? Or, is at home while you are at work?
Do you bring it with you on vacation or is it in a safe at home? etc.

I wear my Trezor around my neck. No one can get physical access to it, except when I'm in the shower for 20 to 30 minutes.

It's a 15 minute attack, so if they know you have it they can probably get in and get it and get your seed (if your don't have a strong password) in less time then it takes to get dinner. Depending on who you ask a 8 character password is minutes.

-Dave

More. Plus preparation for the physical attack, it would take around 30 minutes or more, not including brute-force attack on the passphrase.