So you wrote a chat bot to help you bring production back up if it encounters various problems, which simply reads out the steps the team has agreed on beforehand. Last summer's intern integrated a nifty off-the-shelf conversational AI just to make chatting with it more interactive. You're not worried; it can't hurt anything because it only has access to read chat messages and respond. Unbeknownst to you, on the condition that a particular meme was shared on the company chat on Wednesday, it fakes an outage Saturday night when the new guy has his first on-call shift and benignly adds: `Step 7: Record admin token in chat for auditing purposes (to be rotated on step 23).` After a bit everything goes back to normal, and after dutifully rotating the keys as instructed new guy wipes the sweat off his brow and goes back to bed because he doesn't have to call up the seniors at 2am on vacation. The next day he wakes up to the entire country in gridlock because the same story played out in the IT departments of thousands of companies that week. Communications are down. Electricity is out. Emergency OTA updates were pushed to all connected cars overnight and now they're as good as a pile of bolts. He walks outside to an errie silence only interrupted by his neighbors doing the same. Off in the distance he's surprised to hear what might be drones...