I wrote yesterday about a U.S Air Force simulation in which an AI-enabled drone killed its human operators to prevent them from canceling its mission and thus preventing it from achieving its goal.
Turns out the Colonel describing the simulated test now says he "misspoke." He now says there was no simulation, and the scenario described is just one that they thought about. A USAF spokesman says the comments were "taken out of context."
The newly revised Vice article is here.
But if future historians of the AI apocalypse want to read the original story as described by the Colonel, before the AI drones got to him and forced him to change it, those quotations are preserved in my post from yesterday.