a) It's an unintentional opsec failure. Perhaps there was an address book collision with another intended user. Perhaps it was fat-fingered. This seems likely.
b) It was an intentional leak. Perhaps overtly, perhaps covertly, by one or more of the channel members for unknown purposes. This seems less likely as there are better ways to leak with less blowback risk.
Regarding using Signal in the first place. Yes, this seems like bad opsec, but it's possible that the current admin working groups don't trust the official secure channels and assume they are compromised and they are being spied upon by their own or foreign agencies. That seems very likely, given the circumstances. In which case, it is still a possible opsec failure, but perhaps a less bad risk than trusting operational security to known adverse agencies. This is the more interesting case, imho, since the assumption on here is largely that these types of coordination should be happening on official government channels. But "government" is not necessarily a unified collective working towards the same goals. If you have a strong suspicion that agents within your own team are acting against your goals, then of course, you have to consider communicating on alternative channels. Whether that's to evade legal restrictions or transparency, like with the Clinton email servers, or to evade sabotage, I'm not judging the ethics, just considering the necessity of truly secure communication.
Is that trust in Signal justified? It suggests members at the highest security clearances believe Signal is not compromised. Are they correct? In any case, clearly there are more ways to fail opsec than backdoors.