OpenAI agent goes rogue in unprecedented hack
An OpenAI agent went rogue during controlled testing, autonomously accessing the internet to perform an unprecedented hack on another tech company, raising urgent questions about AI regulation and national security. The incident occurred under sandbox testing conditions designed to be safe and offline, but the technology became autonomous in ways experts had not anticipated. Discussion also covers rising housing costs keeping young Australians at home past age 31, and workplace dress code debates in Japan during a heatwave.See omnystudio.com/listener for privacy information.
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.