RIPA s 12(7)

[Peter Fairbrother]

On 16/06/14 08:04, Caspar Bowden (lists) wrote:
> On 06/16/14 00:26, Peter Fairbrother wrote:
>> On 12/06/14 12:20, Caspar Bowden (lists) wrote:
[...]
>>> Wonder opinions if this sufficient for UK to (coercively) "do a
>>> Hushmail" ? Or under Intel Services Act, or RIPA Pt.2 ?
>>
>> I'm not sure what you mean here.
>
> http://www.wired.com/2007/11/encrypted-e-mai/
>
> Actually I had forgotten that this case involved server-side extraction
> of key (read above). This is obviously within RIP Pt.3 - I remain
> worried about trying to find combo of UK powers which could coerce a
> client-side attack (e.g. he provider has to inject back-doored
> javascript code)

It seems Hushmail had/have two different products, one a Java-based 
applet, one JavaScript-based. The Java-based applet may or may not be 
secure against warrants, but the JavaScript-based one most definitely isn't.

Afaict, in the JavaScript-based app user's passwords and emails are sent 
behind a TSL connection, but otherwise in plain language, nothing is 
hidden from the server - in the case of a sent email, the server then 
does the public key encryption for the recipient. In the case of a 
received email, the server does the private key decryption, then just 
re-encrypts for link TLS.

This of course breaks the end-to-end model, and it is no wonder that 
Hushmail could provide plaintext under a warrant - server-side 
extraction of key, or just supplying users passwords (if that is what 
they did, I'm not clear on that) is just one of many ways in which they 
could have made plaintext available.

I don't see any reason why you couldn't operate a real end-to-end 
encryption scheme via a browser and JavaScript - it would be a bit 
clunky, but I can't see why it wouldn't technically be possible.



So, to get to the question - supposing you did have such a JavaScript on 
your server, could UK Plod force you to backdoor it so you/they could 
read traffic?

I think it depends a bit on the situation, eg what you are protecting, 
email-type messaging or voip traffic might be different - and it might 
also matter whether you are supplying a total service, or just the 
JavaScript software.

On the one hand, if you are just supplying software (and not a telecomms 
service) then I don't think they could force you to backdoor the 
software - on the other hand, if you are supplying a complete voip 
service like Skype (in the UK) then I think they probably could require 
you to have the capability to read traffic on 1 in 10,000 conversations 
- after all, you can't buy telephone equipment which doesn't have 
interception capability these days - even if that requirement meant 
installing a backdoor.

Though the lack of non-interception-capable telephone equipment is 
actually because there is no market for it, rather than any prohibition 
of sales.

In between these, I'm not so sure. May post more later.



-- Peter Fairbrother