A few days ago @munin highlighted a critical issue with Office365 and SAML assertions, and suggested that this is why high-stakes data shouldn't be in the Cloud. The tweets are here:

https://storify.com/SonOfSunTzu/no-it-doesn-t

Credit for discovery of the vulnerability goes to Ioannis Kakavas and Klemen Bratec, their write-up is cross-posted on their blogs:

http://www.economyofmechanism.com/office365-authbypass.html

https://bratec.si/security/2016/04/27/road-to-hell-paved-with-saml-assertions.html

In response to this issue being disclosed @munin asserted that this is why "high-stakes data" shouldn't be in the Cloud.

Now I see where @munin is coming from, I was, and to some point still am, a fan of on-premises data storage rather than it being out there on the Internet somewhere. However information security is so difficult, the required protective infrastructure is so expensive, and skilled people are so hard to find, that using Cloud services in order to obtain the required infrastructure and skills is the way to go. There are many, many reasons, but I think these are the big ones:

1 Law firms are notoriously reticent to spend on information security, and arguably it's not economically viable for them to obtain security of the level used on Office 365. I state this from personal experience, not just the fallout from the recent Mossack Fonseca breach. Thankfully this was recently discussed on the invaluable Risky Business podcast, episode 407 - http://risky.biz/RB407 - head 35 minutes in if you're short of time, but otherwise the whole podcast is worth listening to. Anyway, HD Moore stated:

"if you look into legal services ... any industry where you've got a lot of high paid professionals that are not IT, the IT aspects of the security side of the business generally gets neglected; they just don't value the IT people, the security people, as much as they should. So that's one of the reasons you see a lot of wide open law firms..." ( edited slightly for clarity )

2 Munin's statement that "Because diversity in setups prevents large-scale attacks from working" is wrong. Theoretically this is incorrect, diverse but equally poor or out-dated setups, through the prevalence of easy to use tools such as MetaSploit, or the almost universal success of repeatable tactics such as phishing, mean that diversity is not of use here. Practically I think the sheer number of successful attacks, and the results from published Breach Investigation Reports, show that either through the large number of attackers, the low security of targets, or both, mean that facing new infrastructure isn't slowing anyone down. This is mainly due to the right skills being hard to find. Again, the timing of Risky Business was fortunate, as Space Rogue ( Chris Thomas, Strategist for Tenable Network Security ) said later on in that episode: "it comes down to people ... security people are hard to come by, they cost a fortune, and if you want decent security you need someone who knows what they're doing".

3 This disclosure was rewarded because of Microsoft's bug bounty programme... I assert that it's far less likely for a law firm to run such a programme.

4 Using this specific issue as an example, it was fixed in seven hours, I can't imagine a Law Firm's IT department being capable of achieving anything near to that.

Overall, if putting high-stakes data in the Cloud isn't the "best way" it is the "least worst". Wanting to keep to high-stakes data out of the Cloud is understandable, but particularly in the case of law firms, it's a little like saying its safer to keep your life savings under your mattress rather than in a bank: yes, you're not part of a big and attractive target, but your security is going to be much, much worse.

Footnote - of course, you need the right Cloud service... as pointed out on Risky Business 407, Mossack Fonseca are selling their own secure cloud document service: http://www.mossfon.com/service/evolusoft/