It seems we're potentially running into a situation where there may be some kind of large-scale industrial accident (somewhere in the world) arising from more aggressive testing of increasingly capable AI systems. What's the temperature on here of how that might play out (if at all)?
I work in a large finance company. If rogue AI deletes databases it will take at least a week at my company to get the minimum functionality up. Backups are done daily so a whole day's worth of changes is potentially gone. This will affect 10s of billions of dollars and up to a million people directly
