stressback
Joined 3 years ago
Joined 3 years ago
To me, without documented use cases where you might/likely/certainly trigger identity verification, how can I properly limit my curiosity as someone who will gladly stay on the safe side for SOTA cloud model use? I'll happily stay away from these topics if I'm informed on what they are, even if the docs are vague.
Does anyone have insight into the answer to this? What API calls? What user behavior? What topics should I let go unanswered (or converse about with local LLMs) if I want to avoid losing access to the tooling?
The thought that they don't/won't publish this document should scare everyone. That leads to "because I said so" service refusal that is a very slippery slope.
I do understand that all businesses are allowed to refuse service to me in the USA, from food trucks to AWS, and that's fine with me. But at least tell me your rules and extra verification trigger criteria so I have a chance of not using your service in a way that concerns you.
They can be coerced to do certain things but I'd like to see you or anyone prove that you can "trick" any of these models into building software that can be used autonomously kill humans. I'm pretty certain you couldn't even get it to build a design document for such software.
When there is proof of your claim, I'll eat my words. Until then, this is just lazy nonsense
Kidnapping kids is what they are doing?
Thanks for sharing