|
|
|
|
|
|
by ang_cire
990 days ago
|
|
|
You are missing that the AI is the one creating the output. If I sell you a hammer and nails, I'm not liable if you create a dangerous building. If you ask me to build you a dangerous building and I do it, I am liable if people get hurt. OpenAI wants to pretend that its users are creating the output because they write the prompt, but this is just plainly false, and OpenAI's own limits they put on output shows they know this. Otherwise they'd let the models output information about how to write exploits, how to kill people, etc, which they don't. |
|
|