Four AI lab breaches were caused by a single underlying issue, Irregular says

I thought this was interesting because in the media, it read like the instances involving AI escaping from their test set ups were separate, and the frequency was increasing. This article suggests otherwise, or at least that quite a few instances are actually more this one and the same, just affecting four AI labs at the same time.

So, the models were inadvertently jailbroken in an Ender's Game type scenario.