Just as a word of warning (and a consideration to the "The Q" comments, of which this does reek of advanced mil-int), AES is NIST approved, which in my book always constitutes it as being 'backdoored'.
Certain types of AES (such as NoPadding variants or ECB - Electronic Code Book) have fundamental weaknesses (NoPadding is vulnerable to 'Oracle attacks' - look it up), and my thoughts are the NSA (who likely nudged NIST into approving AES - remember, AES is used by the US government) built fundamental mathematical backdoors into it (basically, a mathematical backdoor is 'if you calculate X in a slightly different way, you get more efficient returns, EG in brute forcing attacks').
What 'StarGate' is describing sounds EXACTLY like a mathematical backdoor (the whole point of an encryption algorithm is that it's meant NOT to be 'easily brute forceable' if you change certain aspects of it), and I highly doubt they just 'casually' discovered such a backdoor in an algorithm that has been around for quite some time, which either means:
1) SG is some sort of insane genius (timing is circumspect, as is their disregard for Julian Assange's safety - they could have kept this quiet), but with no prior history establishing as such I doubt this.
2) SG works for a research body or university with the resources necessary to make solving this equation possible (IE with cryptographic experts, mathematical professors, and the like), or the most likely in my book,
3) SG is an NSA employee and the NSA, in their sheer desperation and panic over Q, and in an effort to neutralise Julian Assange in the same move, have voluntarily given up a mathematical backdoor in their own proposed encryption (note: this is a double-edged sword, ANY AES based encryption is hypothetically exposed - although I notice SG is VERY quiet on the details of how exactly he did it) and are thus publishing the passcodes in order to nullify Julian Assange's insurance policy.
If it is the third case, then we can be confident the NSA already read the insurance files, and have concluded either there was nothing of merit, or they've already solved whatever problems they've presented.
It's unclear what benefit it would have to expose this information publicly though (why not just neutralise Julian Assange and let the whole thing naturally flop?), unless their goal is to now discredit Julian Assange by basically saying 'hah, his insurance files contain nothing (because we've already purged the evidence'.
Either way, something about this reeks, and I'd advise we dig just a little further into StarGate to validate their claims. People don't just make major breakthroughs on cryptography overnight - usually specialist researchers 'chisel away' at a problem set.