the one benefit i see is as a final “lock in your answer” function for any form you’re submitting.
And did they have to stop allowing written signatures because of AI scraping? Or is it just because they never looked legible at all?
the one benefit i see is as a final “lock in your answer” function for any form you’re submitting.
And did they have to stop allowing written signatures because of AI scraping? Or is it just because they never looked legible at all?
The visual signature part of a “digital signature” is purely cosmetic. The actual signature is the entire file being cryptographically signed using a private key you possess, which any recipient can then verify came from you by using a public key.
This is the case for all government issue signatures (like when using the private key inside a chipped government id card) or any other signature created using a key pair from an issuing authority.
If its not a digital signature in this way, the other way is literally just an image. It can be a vector or an raster, or a straight up scan of a physically signed document. It’s just a picture. It has no digital verification whatsoever.
I’ve done that a few times, in GIMP (or Photoshop), and was always a little surprised that it seemed to be enough. Maybe I unwittingly “forged” my own signature? That’s legally murky…
How is it legally murky? Someone asks if that’s your signature on the document and do you remember signing it, and it is and you do, so you say yes. End of discussion.
It’s only a problem if you’re claiming it’s not your signature. But then, they have to prove it’s your signature, you don’t have to prove it’s not. I’m not sure you can prove it’s not your signature, hence burden of proof being on the other party.
It’s really just a digital way to do the equivalent of printing the document, signing it, and scanning it again. Which is also accepted in a lot of situations.
Ive signed many contracts like that, never had an issue
Surely they’re talking about the likes of DocuSign, which doesn’t involve any user-controlled keys
The only private key with docusign is the fact you has access to the email inbox. And it’s not even a private key, just a symmetric key in the link of the email. They might do some basic user agent and IP checks for suspicious activity, but if someone you know has access to your email DocuSign wouldn’t know the difference
🤯
Well that explains everything. Thank you very much!
That didn’t quite explain everything. I want to add that there are “type your name to sign” forms that are just what you described, just bytes spelling out your name. In those cases, you’re right, those bytes don’t have your handwriting to prove anything, but they do accomplish something. If the form is secured behind a login and not a totally buggy heap of garbage, then the act of typing your name on that form demonstrates that whoever logged in deliberately expressed their approval. And whatever level of identity verification is needed for login is then playing the part of demonstrating who did that. That’s not perfect, but the level of evidence might match the level of risk enough to justify that mechanism to the company providing it, and it’s important to consider all factors: not just system cost, but end user tolerance.
E.g., there was (maybe still is?) a “ham” radio logging system that had end users generate their own PGP signing key pair to sign their log books, which are simply a list of who they talked to over the radio, always for non-commercial purposes (by law). Many questioned whether a secure web site with good 2FA might suffice and be much more convenient, manageable, and adequate risk level for most users.
“Digital signature” is a proper term that should mean what the parent comment explained–public key crypto “signing” (a hash of) a document’s content–but “sign here by typing your name” will inevitably get called a signature and, obviously, is also digital, just not a real “digital signature”.
Thank you, I was definitely thinking of the first paragraph but the context of third paragraph (and MentalEdge’s comment) helps me to understand why that system actually works/is considered good enough.
In the US, they don’t do the cryptographic signatures, they do the image. There are document signing services where you just type your name and it creates a “handwritten” signature from that name and that’s legally binding apparently.
I found it very funny when I had to sign a contract with an American company that way. Not sure it would’ve been legally binding in my jurisdiction, but the contracts was more or less “don’t tell anyone about exact details of the sizable bonus you’re about to receive, or we’ll revoke it, also you’ll have to stick around for half a year to receive the other half”, when the company got sold to private equity and the original owners decided to give us all something to remember them by.
Old school contract law was always about expressed mutual understanding, and predates widespread ability to read and write so didn’t require papers (but did like having witnesses of the agreement).
You still see surviving requirements of some contract signings needing witnesses.
It’s still the same in the digital age, if you can demonstrate all participants agreed then the form doesn’t matter much (aside from in regulated fields, some have their own requirements on form).
What? No it’s not. This whole comment is wrong, sorry.
When a website has you type a name as a signature it’s a legal representation of that signature. There’s nothing cryptographic at all about it. In the US this is regulated by the ESIGN Act (https://uscode.house.gov/view.xhtml?path=%2Fprelim%40title15%2Fchapter96&edition=prelim). Also, where would you even store the key? I’ve made ways to use Webauthn keys to do this, but no signatory uses these methods.
And that’s not how ZKPs in government IDs work, either. You never sign with them, you only verify.
Ref: My job is cryptographic signature protocols, and I have a draft at IETF for one of them.
I claimed no such thing. That is not a digital signature.
I’m fairly certain we are not discussing the same thing.
I can digitally sign documents the way I have described, NOT IN A WEB BROWSER, using entirely local or even open source software, using my government issued ID card.
As described by the relevant finnish agency, this is enabled by a “signature certificate” stored on peoples ID cards. The functionality uses a separate PIN from that which you use to verify your identity in the context of a website you are visiting.
I am not aware of any other way for me to sign a document in such a way. All the other signatures, which do also work, are something else. And potentially insecure in all the ways OP thinks. (Someone else can type the same name).
But that’s what OP is describing with the “”'s, so I assumed you were responding to the question, which they may have as well.
Interesting, I hadn’t realized Finland added a separate cert from the ZKP ( https://www.suomi.fi/services/citizen-certificate-digital-and-population-data-services-agency/dc540ff4-0030-46b2-add0-9f7ceb2a41c8), new info gleaned.
Though this may still be “slightly” off. Seems the finish ID is a standard Zero Knowledge Proof (ZKP) plus a pin activated signature (Citizen) certificate.
This should mean you can identify yourself, or that you are over 18 without revealing your full identity) without the PIN, as that operates separately from the signing certificate which can be activated independently. The PIN would be used for signatures or encrypting things. Very cool.
There are two PINs, PIN1 for ZKP and PIN2 for digital signing. The card does nothing without one.
Most EU countries provide some way for citizens to sign documents using a private key tied to their legal personage.
Putting it on the ID card they are given anyway just makes sense.
My final paragraph is there to explain that a lot of “digital signatures” aren’t really signatures in any meaningful way. What goes into the actual file doesn’t matter at all, UNLESS it’s the aforementioned cryptographic signature.