The AI companies are apparently inundated with flags of all sorts of behaviour from criminal to oddly sexual etc prompts from 160+ countries around the world. I don’t think their terms of service guarantee anything around safety specifically. My point really though was, whilst these companies should put the guardrails in place, the primary blame still sits with the human who made the decision to act illegally. We should not be turning away from the actual actor. Humans are also using devious ways to get around the guardrails (like Tesla drivers who do hacks to bypass safeguards for autopilot), and these are deliberate actions.

Can’t remove or deflect blame from the actual perpetrator. AI is often used for good. My wife used it to learn how to start a business and the accounting system. AI is like a car, a public road, a kitchen knife, a gun, etc - it can be used for many purposes, but it is who uses it for what, which differs. Same goes for AI hacking other companies. The blame sits squarely with that AI’s owner who still had it connected to the Internet. AI is like a toddler, it is not intelligent, or even close to superintelligent, it is artificial.