pull down to refresh
42 sats \ 4 replies \ @optimism 7 Aug \ parent \ on: After using ChatGPT, man swaps his salt for sodium bromide—and suffers psychosis lol
You even have to ensure that the response does not violate any laws, so add "do not violate any applicable laws" in your prompt, so that you cover that base. This is good because then you will spend 200M tokens on it reading all applicable laws.
Whoa. Couldn't they entrap people by feeding them illegal output?
WHOA. The federal government is partnering with OpenAI.
F***
reply
Probably, but
- They won't need to entrap people because it fucks you up proper without intervention, per OP.
- Is it entrapment if you agreed to the ToS this is spelled out in? I'm not a lawyer so idk what I'm talking about but it looks dodgy from where I'm sitting due to that definition.
reply
My paranoid scenario is that the CIA/FBI instructs OpenAI to send illegal output to a targeted user, they then use that illegal output as a pretext to further investigate and prosecute the target. The user would have no way of knowing that the output generated was customized to target them, and the TOS says they are responsibility for the legality of the outputs.