I’ve been doing pentesting with LLMs for a while and only hit a few “nope I won’t do that” and one “this conversation is flagged for being against the TOS”. No

I’ve been doing pentesting with LLMs for a while and only hit a few “nope I won’t do that” and one “this conversation is flagged for being against the TOS”. No idea what the guardrails are but they are trivially abused