I think you better tell users to be careful of case for the secure code.
or maybe just eliminate case and make the code longer.
I had considered removing case and making it longer, but I really wanted it to be shorter to avoid having it significantly increase the amount of data to write/type. I already think it's too long, but I need to make sure it has some kind of sanity check, and enough entropy to be useful.
On that note: at the moment, it's 7 bytes plus a 1-byte checksum displayed in base58. That is 56 bits of entropy, drowned in 16MB of key-stretching. 56-bits doesn't sound like a lot, but the keystretching takes my i5-2500K about 0.25seconds to compute. If you had just a single CPU working on that, it would take 580 million years to go through the possible keyspace. Even with a multi-million-CPU botnet, you're out of luck. And GPUs won't be very useful with the 16MB required per thread.
I figured the case-sensitivity was okay, because I'm displaying it in such a huge font, it's obvious what is upper and lower case. It can still be changed, though...
Also, I would remove characters like 1, I, i, 0, O, o etc. to prevent confusion.
Btw really looking forward to the M-of-N function, with this I can create my final, long-term Horcruxes. I'm thinking of adding one to my testament, since that document will probably be around pretty much forever (I believe there is a national archive of some sorts for these things). Thanks for all the great work!