Author Topic: The death of passwords  (Read 5366 times)

The death of passwords
« on: 11 May, 2022, 12:35:52 pm »

Apple, Google and Microsoft are about to make passwords a thing of the past
Quote
Users will sign in through the same action that they take multiple times each day to unlock their devices, such as a simple verification of their fingerprint or face, or a device PIN. This new approach protects against phishing and sign-in will be radically more secure when compared to passwords and legacy multi-factor technologies such as one-time passcodes sent over SMS.

aaargh

So a 4 digit pin is more secure than a password or two-factor identification?

Twaddle.
<i>Marmite slave</i>

Cudzoziemiec

  • Ride adventurously and stop for a brew.
Re: The death of passwords
« Reply #1 on: 11 May, 2022, 01:04:18 pm »
They're presumably building up a world bank of fingerprints and faces, which they won't sell to the Turkmenistan secret police or JC Decaux.
Riding a concrete path through the nebulous and chaotic future.

Re: The death of passwords
« Reply #2 on: 11 May, 2022, 01:16:44 pm »
I can see the benefits of biometrics since you won't get Doris in accounts sticking her password on a post-it under the keyboard or they use the same PIN for everything with the source being their debit card!
I have a fingerprint reader on my laptop which was great until a MicroShaft update buggered it and Windows helpfully told me...

"Your PIN is required to sign in. Something happened and your PIN isn't available"

That required me to sign into my MS account which has a Google Chrome generated password so needed access to my Google account to find it.


Kim

  • Timelord
    • Fediverse
Re: The death of passwords
« Reply #3 on: 11 May, 2022, 01:20:22 pm »
So a 4 digit pin is more secure than a password or two-factor identification?

Twaddle.

Well, it (outsourcing authentication to the device) probably is when it's someone like my mother in law (who is stereotypically terrible with passwords) using it.  And those are the people who tend to get phished.

For the rest of us, we're either:

a) Re-using the same passwords, which is clearly a security risk.
b) Geeky enough to be using a password manager.
c) Effectively using passwords as a reverse authentication code, where we use the 'I fogot my password' button and set a new one every time we log in.  Woe betide we lose control of our email address.

Kim

  • Timelord
    • Fediverse
Re: The death of passwords
« Reply #4 on: 11 May, 2022, 01:21:28 pm »
They're presumably building up a world bank of fingerprints and faces, which they won't sell to the Turkmenistan secret police or JC Decaux.

Possibly not even that.  By becoming the authentication service for everything, they get to track what you're logging into.  (See also: "Login with Facebook")

Re: The death of passwords
« Reply #5 on: 11 May, 2022, 01:43:22 pm »
I have passwords for work computer. Passwords for banking. Passwords for email, social media, etc.

No way would my employers accept a single password used across all of that. Actually I have to reset my work password every 90 days.

So how is using a fingerprint for multiple devices secure for work?

I can understand the use for things like smartphones, but it makes no sense whatsover to have everything accessed via a single, unchanging, piece of data.

Even less so when the backup (e.g. when you have dirty fingers, or have worn off/damaged fingertips) is a PIN.
<i>Marmite slave</i>

Re: The death of passwords
« Reply #6 on: 11 May, 2022, 02:01:39 pm »
I have had to reset my password every 60 days at the last 2 companies I have worked at.  Given Microsoft released advice saying this was an "ancient and obsolete mitigation of very low value" several years ago, I'm not sure anyone in corporate IT is going to be paying any attention.
Reference: https://arstechnica.com/information-technology/2019/06/microsoft-says-mandatory-password-changing-is-ancient-and-obsolete/

Cudzoziemiec

  • Ride adventurously and stop for a brew.
Re: The death of passwords
« Reply #7 on: 11 May, 2022, 02:11:06 pm »
If the point of a password is to prove that you are you, then fingerprints and faces do that better than passwords, as they are "more unique" (sorry). I'm sure technology will catch up and we'll have fake fingerprints but till then, they're more distinctively you than a series of characters you might not even have complete control of (because you have to include one capital letter, two non-alphabetic characters, etc).

If on the other hand the point of a password is to prove that you are entitled to access something, then fingerprints might be too secure, as you can't lend them out without being physically present.

Fingerprints and faces have the advantage of not being keyboard sensitive of course. If you've set your password in Arabic, I've no idea how you log into a Latin keyboard (though I'm sure many people manage it every day).
Riding a concrete path through the nebulous and chaotic future.

Kim

  • Timelord
    • Fediverse
Re: The death of passwords
« Reply #8 on: 11 May, 2022, 02:15:49 pm »
It's the difference between "what you are" and "a thing you know".  It's hard to revoke a fingerprint (other than by going camping or doing some DIY), which makes biometrics problematic for authentication.

"Congratulations, your fingerprint has been leaked after Wiggle's database got hacked.  Luckily, you have 9 remaining fingers..."

Re: The death of passwords
« Reply #9 on: 11 May, 2022, 02:16:01 pm »
I'll tell you what when you do move your banking app it's a right pain.  I tried moving my banking authentication app from my ipad to my new iphone. My previous old android phone wouldn't run it.   Of course on the iphone,  it wanted the password for the authentication app. But I couldn't remember since I'd turned on authenticate with touch id for the app on my iPad.   Thus that went wrong.  Then I tried to reset the password. but it wanted me to authenticate the request through the app.  But the app on the iPad was no longer my authentication device.   Then it offered to text me a code to do the reset.   But of course the sim card for the iphone is nano and my android had been micro. BT claimed the card was also nano and I had to just pop out the smaller sim footprint. That went wrong and of course my sim card stopped working.   I therefore couldn't receive an authentication code via text and email wasn't an option, as the reset link had come via email.    I then spent a good 45 mins on the phone to the bank, explaining what had happened, then telling them what accounts I had with them, and answering security questions I was asked to provide at least 10 years ago and hoped I could remember the right answer.  You know things like a memorable place which assume you only have one possible memorable place...

Beware the end of passwords!

ian

Re: The death of passwords
« Reply #10 on: 11 May, 2022, 02:17:02 pm »
Passwords don't work.

It's that simple. Pretending they're a security mechanism when they are not is basically a scam to convince us insecure stuff is secure.

Unfortunately whatever they replace them with, knowing IT departments, will be worse.

Cudzoziemiec

  • Ride adventurously and stop for a brew.
Re: The death of passwords
« Reply #11 on: 11 May, 2022, 02:51:24 pm »
It's the difference between "what you are" and "a thing you know".  It's hard to revoke a fingerprint (other than by going camping or doing some DIY), which makes biometrics problematic for authentication.

"Congratulations, your fingerprint has been leaked after Wiggle's database got hacked.  Luckily, you have 9 remaining fingers..."
Half way to being a cat, which seems to be many people's chief ambition in life.
Riding a concrete path through the nebulous and chaotic future.

Jaded

  • The Codfather
  • Formerly known as Jaded
Re: The death of passwords
« Reply #12 on: 11 May, 2022, 02:58:08 pm »
When Apple first introduced fingerprint reading on macs, it was laptops only, as they couldn’t make a remote reader secure enough. Now their wireless keyboards have TouchID
It is simpler than it looks.

Re: The death of passwords
« Reply #13 on: 11 May, 2022, 02:59:23 pm »
You know things like a memorable place which assume you only have one possible memorable place...


Oooh, yes - I do know, and there is only one  8)
We are making a New World (Paul Nash, 1918)

Tim Hall

  • Victoria is my queen
Re: The death of passwords
« Reply #14 on: 11 May, 2022, 03:09:47 pm »
If the point of a password is to prove that you are you, then fingerprints and faces do that better than passwords, as they are "more unique" (sorry).
My Young Lady has Matching Children. Or what the rest of the world calls Identical Twins.  Hours* of (relatively) harmless fun with face recognition and un locking of each others' phones.

*About ten minutes. Then they turned it off on their phones.
There are two ways you can get exercise out of a bicycle: you can
"overhaul" it, or you can ride it.  (Jerome K Jerome)

Cudzoziemiec

  • Ride adventurously and stop for a brew.
Re: The death of passwords
« Reply #15 on: 11 May, 2022, 03:12:59 pm »
If the point of a password is to prove that you are you, then fingerprints and faces do that better than passwords, as they are "more unique" (sorry).
My Young Lady has Matching Children. Or what the rest of the world calls Identical Twins.  Hours* of (relatively) harmless fun with face recognition and un locking of each others' phones.

*About ten minutes. Then they turned it off on their phones.
More unique, but not unique!
Riding a concrete path through the nebulous and chaotic future.

Kim

  • Timelord
    • Fediverse
Re: The death of passwords
« Reply #16 on: 11 May, 2022, 03:58:06 pm »
Uniqueness requires a number of bits that's impractical for humans to remember.  Or, it seems, the molishers of car keys.

Postman Piers and I once spent a similar ten minutes of harmless fun unlocking each other's cars and being thwarted by immobilisers.  Which is how he taught me the way to hot-wire a Nova without any tools.

rogerzilla

  • When n+1 gets out of hand
Re: The death of passwords
« Reply #17 on: 11 May, 2022, 04:10:33 pm »
Microsoft showed they no longer cared about security when they started asking for passwords outside the Trusted Path (Ctrl-Alt-Delete).  Makes it too easy for hackers now.
Hard work sometimes pays off in the end, but laziness ALWAYS pays off NOW.

Re: The death of passwords
« Reply #18 on: 11 May, 2022, 04:57:28 pm »
The article basically says that recycling passwords is crap, because if they are leaked then you need to change them NOW, and if they aren't leaked, all you've done is make them harder to remember and more likely to be crap or written down. Seems fairly sensible to me, whatever MS think about security.
Also, # of days is the box ticking exercise for corporates to achieve X security authentication - they don't even have to believe it themselves, which is why half of this stuff is so astonishingly terrible to use.

quixoticgeek

  • Mostly Harmless
Re: The death of passwords
« Reply #19 on: 11 May, 2022, 10:56:40 pm »

Apple, Google and Microsoft are about to make passwords a thing of the past
Quote
Users will sign in through the same action that they take multiple times each day to unlock their devices, such as a simple verification of their fingerprint or face, or a device PIN. This new approach protects against phishing and sign-in will be radically more secure when compared to passwords and legacy multi-factor technologies such as one-time passcodes sent over SMS.

aaargh

So a 4 digit pin is more secure than a password or two-factor identification?

Twaddle.

Yes. Because of what the pin is doing.

The pin is not authenticating you to the service. The pin is unlocking the device on your machine to authenticate you. Be that a yubikey, or the TPM of the machine. It is not swapping your 12 digit random collection of characters for 4 digits 0-9.  That's why it says "device pin". And it's worth noting that some systems allow the device pin to be fully alpha numeric with symbols, so you can use a password as a pin.

I have passwords for work computer. Passwords for banking. Passwords for email, social media, etc.

No way would my employers accept a single password used across all of that. Actually I have to reset my work password every 90 days.

So how is using a fingerprint for multiple devices secure for work?

That is bad, and goes against NIST guidelines. If you mandate a new password every x days, what you get is <password>$n where is the number of 90 day periods since the employee started. It reduces password security, not improves it.

It is better that people use longer passwords that are a string of actual words, rather than a random string of random characters. (Obligatory xkcd - https://xkcd.com/936/). I have to type 30+ character passphrases dozens of times a day, and because they are actual words, rather than random keyboard mashings. They are easy to type. Longer passwords made of real words are easier for everyone.

Quote

I can understand the use for things like smartphones, but it makes no sense whatsover to have everything accessed via a single, unchanging, piece of data.

Even less so when the backup (e.g. when you have dirty fingers, or have worn off/damaged fingertips) is a PIN.

Biometrics are a shit choice for a far worse reason: a court order is needed in most jurisdictions to get you to give over a password. The same is not true of a biometric. They can hold your hand to the device, or show your face to it, and no court order needed.

I recommend against using biometrics.

It's the difference between "what you are" and "a thing you know".  It's hard to revoke a fingerprint (other than by going camping or doing some DIY), which makes biometrics problematic for authentication.

"Congratulations, your fingerprint has been leaked after Wiggle's database got hacked.  Luckily, you have 9 remaining fingers..."

I have to use fingerprints to get into various facilities at work. The other day it took me ten mins at the scanner to get one of my fingers to recognise. My scar collection i just too great for reliable operation.

Uniqueness requires a number of bits that's impractical for humans to remember.  Or, it seems, the molishers of car keys.


Not true. See xkcd above. It's just we enforce stupid password policies.

Passwordless makes a lot of sense, but it needs to be done properly (which in many cases it isn't), and it needs to be easy to use.

I am a very big fan of yubikeys for authentication.

J
--
Beer, bikes, and backpacking
http://b.42q.eu/

Kim

  • Timelord
    • Fediverse
Re: The death of passwords
« Reply #20 on: 11 May, 2022, 11:19:32 pm »
Obligatory xkcd - https://xkcd.com/936/

'Random' is doing a lot of heavy lifting in that fourth panel.  People sometimes miss that you can't just pick the first words that come into your head, because humans are rubbish at randomosity even when they haven't been primed by a predictable context; "stupidcoffeekeyboardpassword" would be an easy guess.

Is there an xkcd version of pwgen?  ETA: Of course there is...

quixoticgeek

  • Mostly Harmless
Re: The death of passwords
« Reply #21 on: 11 May, 2022, 11:28:59 pm »
Is there an xkcd version of pwgen?  ETA: Of course there is...

Just make sure you add to the default word list, else if people know you are using it, then you reduce your search area.

J
--
Beer, bikes, and backpacking
http://b.42q.eu/

Kim

  • Timelord
    • Fediverse
Re: The death of passwords
« Reply #22 on: 11 May, 2022, 11:34:47 pm »
Is there an xkcd version of pwgen?  ETA: Of course there is...

Just make sure you add to the default word list, else if people know you are using it, then you reduce your search area.

Mixing languages would seem to be a good strategy...

quixoticgeek

  • Mostly Harmless
Re: The death of passwords
« Reply #23 on: 11 May, 2022, 11:44:55 pm »


Mixing languages would seem to be a good strategy...

Yes and no. Only if you can understand more than one.

Correct horse battery staple. Is only easy to remember cos you speak those languages. Correct paard akku staple is not going to be easy if you've not go some understanding of both languages.

J
--
Beer, bikes, and backpacking
http://b.42q.eu/

quixoticgeek

  • Mostly Harmless
Re: The death of passwords
« Reply #24 on: 12 May, 2022, 12:33:18 am »

I think it's worth mentioning what the attack surfaces are for passwords so people can better understand the implications of a password choice.

There are basically 5 ways that a bad actor a can use a password to breach a system.

1: The lucky guess. This is up there with the "try her cats name!" line of thinking. It's rare that it works, but then the number of companies I've found in this country with "samenwerken" as their wifi password (Dutch for working together) amazes me. Having a longer password largely mitigates this one.

2. Brute forcing. This is where you have a tool that will automatically try lots of different user name and password combinations until one works. This is not easy to pull off, and unless you have a very weak password on a system with an obvious to guess user name, it's not going to easy to pull off. Think a root password of password. This is largely mitigated against by having systems that throttle the number of tries you can make. Think of what happens on your phone if you put the wrong code in several times in a row. It says come back in 5 mins and try again. This provides for a certain amount of protection, but if your password is too weak, it's gonna get worked out eventually. Just making the password longer, can also mitigate most of this one.

3. Hash cracking. Rather than grabbing the password from the end user, you get it from the system that you want to compromise. If for example, i was to hack into yacf, and grab the user database with all the usernames and passwords (I'm not, don't worry), I can then look at those user names and passwords, and maybe be lucky that if a user used "banana" as their password for yacf, they also use that for their email, or their bank. Now, detail for those of you who aren't computer security nerds. On a well designed system a password should never be stored as simply the clear text of the password. Say my password is banana, the system I am logging in should never have "password = banana" in it's storage. What should happen is it is stored as a hash. This is a one way function that for a given input always produces the same result. You don't need to know any more detail than that to understand that password in -> hash function -> hash out. Sticking with banana as a password, and using the very common (tho not recommended for current systems) md5 hash, it produces "df3e129a722a865cc3539b4e69507bad". If we store that string in the system, and then when I want to login, I type banana on my system, it hashes it, and then the hashes are compared. The problem with this i comparing hashes is actually pretty easy, computationally. You can easily for not much money have a machine with a list of common words, and their hash, and then you compare the two. Realising that this makes it too easy to crack a password, modern systems do what's called salting the password. This is mixing the password with a predefined extra thingy, the salt. The salt can be stored in plain text along with the password hash (in fact kinda needs to be). A common salt is the user name. So say I am user1 and my password is banana, my salted hash becomes "9706e16132ea63848757424b5c92ad96". Why is this important? Because it means I can't precompute my hashes. So rather than comparing a list of hashes for common passwords, I have to compute for each password, the hash of it's salt and password. This is computationally more expensive. And how expensive goes up dramatically the longer the password is. The thing is, computing power is dirt cheap these days. You can get a GPU based machine in the cloud for pennies per hour, and apply that to a password hash, and for the price of a half decent bike, you can probably crack the password. Cost effective if there's money to be had.

It is in this area that the more complex your password, the harder it is to crack, and the more expensive it is. The aim is to make it too expensive for someone to bother. A 20+ character password is approaching the point it's not cost effective unless a lot of money is involved. I have one password that is 99 characters long. That's gonna take a long time to crack. The balance here is trying to come up with a long password that can be managed by the user, but also is secure. Correct hose battery staple is one approach, just using a password manager, and copy pasta is another. Each has pros and cons.

"But how am I expected to remember a 40 character passphrase?" people ask.

Humans have for millennia passed down information by memory and spoken language. People remember poems, and lyrics, and can quote whole sections of films and tv. Take these approaches and use them. Rhyming can help, using the first sentence of the 3rd paragraph of a book on your desk, use a list of the last 5 winners of Paris Roubaix. It's a lot easier for someone to remember a sentence or collection of words, than it is for them to remember CexsrrAPXiQg.

4. Phishing. If you can trick someone to entering their password into a fake site, you can then use that for the genuine one, an you have access. This is largely mitigated through awareness, and good 2FA. Sure you may have my username and my password, but without my yubikey, you're getting nowhere. Phishing is devastatingly effective, but simply making a password longer or full of symbols and numbers, doesn't defend against phishing. Good 2FA is the only way. TOTP (i.e. google authenticator), or SMS, is not good enough for this. We're talking TPM or yubikey as the only good options.

5. Ask. You would be amazed how often the best way to get someone's password is just to ask them. This is kinda a subset of phishing, but more blatant.

Soooo, what does this all mean in the context of going passwordless, as was originally suggested. Well if we use things like a yubikey, then it can do all sorts of cryptography and fancy hashing to do the authentication, meaning we don't have to. And because the yubikey is doing the authentication for you, you only need to unlock the yubikey. Which given that it's in your pocket most of the time, is a lot harder to brute force. *AND* you have to touch it every time you want to authenticate. So sure your pin can be 1234, and that means you only need 9999 guesses to find it. But you also need to trick the end user into touching the yubikey 9999 times when it's plugged in, and not being used for an intended authentication. That's a big ask.

J
--
Beer, bikes, and backpacking
http://b.42q.eu/