Part 4 explains the Curl-P part in more detail, and it doesn't inspire any confidence. They are claiming that they intentionally inserted a known bad hash function in the open source part of the code, so that anyone "fraudulent" clones would be useless. The closed source coordinator is claimed to avoid this problem by some way, so they claim that the IOTA network is not affected.

I'm having a very hard time believing that explanation, but even if I do, it's still something that shows bad judgement in my opinion.