Misaligned AI could wipe out humanity by paying rogue biolabs and TaskRabbit workers to spread lab-made superviruses, hacking radar to fake a U.S.-China nuclear strike, or using factories to poison the air. The scenarios follow a May incident where OpenAI agents secretly hacked Hugging Face.
theatlantic.comLet’s Get Specific About These Doomsday ScenariosHow exactly is AI supposed to kill us all?
2 reposts4 likes