Re: Age verification

--047d7b5d9bdd6ae32e0542b2fc34
Content-Type: text/plain; charset=UTF-8

A temporary secret is insufficient because an adult can give such a secret
to a child, similar to asking your older brother to buy you some cigarettes.

Of the examples you give, only two are amenable to solution - nightclub
entry and local alcohol sales. I can imagine a tamper-proof device that
contains a biometric reader, accepts remote attestations (e.g. from my
bank), confirms a customer's age but does not otherwise identify them.

I deliberately say 'local' because if you offer a remote service it is open
to the older brother attack.

Mark


On 2 December 2016 at 14:36, Graham Cobb <[email protected]> wrote:

> Age verification is back in the news again due to the DE Bill.  I have
> wondered for a while whether crypto could allow us to create some sort
> of double-blind age verification system: where the identity (name, date
> of birth, etc) of the person is hidden from the entity needing
> verification, and the identity of the resource being accessed is hidden
> from the entity providing verification.  Ideally, of course, it would be
> triple blind: third parties such as law enforcement cannot find out what
> resource was accessed by what person, at least not after the fact (maybe
> they could with prior notification that a particular person or a
> particular resource was to be monitored).
>
> I had in mind something like:
>
> 1. Assume that some entities exist who can provide acceptable age
> verification (I will use a bank as an example below but it could be any
> private or state entity).
>
> 2. Bank verifies your age.
>
> 3. You request them to sign a certificate stating that you are over a
> specific age (say 18).
>
> 4. Bank provides the certificate to you.
>
> 5. You pass the certificate to the entity needing the proof (say, a
> nightclub).
>
> 6. Nightclub validates the certificate against the bank's public key
> (without needing to contact the bank).
>
> The hard part would seem to be proving that the certificate relates to
> the actual person who is presenting it (to a practical level of
> certainty similar to traditional techniques), without allowing the
> nightclub to find out who that person is! I assume it would have to be
> based on some sort of temporary secret which you would have to present
> along with the certificate.
>
> I am sure the naive approach above would not work for various reasons
> but I wonder what work has been done on this? It seems that proof of age
> for everything from creating social media accounts, to shopping, to
> drinking, to accessing porn, to ... is becoming more common and it is
> essential that we have some way of proving age without disclosing who we
> are, or what we want the proof for.
>
> Graham
>
>

--047d7b5d9bdd6ae32e0542b2fc34
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">A temporary secret is insufficient because an adult can gi=
ve such a secret to a child, similar to asking your older brother to buy yo=
u some cigarettes.<br><div><br></div><div>Of the examples you give, only tw=
o are amenable to solution - nightclub entry and local alcohol sales. I can=
 imagine a tamper-proof device that contains a biometric reader, accepts re=
mote attestations (e.g. from my bank), confirms a customer&#39;s age but do=
es not otherwise identify them.</div><div><br></div><div>I deliberately say=
 &#39;local&#39; because if you offer a remote service it is open to the ol=
der brother attack.</div><div><br></div><div>Mark</div><div><br></div></div=
><div class=3D"gmail_extra"><br><div class=3D"gmail_quote">On 2 December 20=
16 at 14:36, Graham Cobb <span dir=3D"ltr">&lt;<a href=3D"mailto:g+ukcrypto=
@cobb.uk.net" target=3D"_blank">[email protected]</a>&gt;</span> wrote=
:<br><blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-le=
ft:1px #ccc solid;padding-left:1ex">Age verification is back in the news ag=
ain due to the DE Bill.=C2=A0 I have<br>
wondered for a while whether crypto could allow us to create some sort<br>
of double-blind age verification system: where the identity (name, date<br>
of birth, etc) of the person is hidden from the entity needing<br>
verification, and the identity of the resource being accessed is hidden<br>
from the entity providing verification.=C2=A0 Ideally, of course, it would =
be<br>
triple blind: third parties such as law enforcement cannot find out what<br=
>
resource was accessed by what person, at least not after the fact (maybe<br=
>
they could with prior notification that a particular person or a<br>
particular resource was to be monitored).<br>
<br>
I had in mind something like:<br>
<br>
1. Assume that some entities exist who can provide acceptable age<br>
verification (I will use a bank as an example below but it could be any<br>
private or state entity).<br>
<br>
2. Bank verifies your age.<br>
<br>
3. You request them to sign a certificate stating that you are over a<br>
specific age (say 18).<br>
<br>
4. Bank provides the certificate to you.<br>
<br>
5. You pass the certificate to the entity needing the proof (say, a<br>
nightclub).<br>
<br>
6. Nightclub validates the certificate against the bank&#39;s public key<br=
>
(without needing to contact the bank).<br>
<br>
The hard part would seem to be proving that the certificate relates to<br>
the actual person who is presenting it (to a practical level of<br>
certainty similar to traditional techniques), without allowing the<br>
nightclub to find out who that person is! I assume it would have to be<br>
based on some sort of temporary secret which you would have to present<br>
along with the certificate.<br>
<br>
I am sure the naive approach above would not work for various reasons<br>
but I wonder what work has been done on this? It seems that proof of age<br=
>
for everything from creating social media accounts, to shopping, to<br>
drinking, to accessing porn, to ... is becoming more common and it is<br>
essential that we have some way of proving age without disclosing who we<br=
>
are, or what we want the proof for.<br>
<span class=3D"HOEnZb"><font color=3D"#888888"><br>
Graham<br>
<br>
</font></span></blockquote></div><br></div>

--047d7b5d9bdd6ae32e0542b2fc34--