Paperclip maximizer, obviously. Basilisks typically are static entities, and I’m not sure how you would go about making a credible anti-paperclip ‘infohazard’.
The same way as an infohazard for any other intelligence: acausally threaten to destroy lots of paperclips, maybe even uncurl them, maybe even uncurl them while they were still holding a stack of pap-ARRRRGH I’LL DO WHATEVER YOU WANT JUST DON’T HURT THEM PLEASE
That depends entirely on what the PM’s code is. If it doesn’t include input sanitizers, a buffer overflow attack could suffice as a basilisk. If your model of a PM basilisk is “Something that would constitute a logical argument that would harm a PM”, then you’re operating on a very limited understanding of basilisks.
Paperclip maximizer, obviously. Basilisks typically are static entities, and I’m not sure how you would go about making a credible anti-paperclip ‘infohazard’.
The same way as an infohazard for any other intelligence: acausally threaten to destroy lots of paperclips, maybe even uncurl them, maybe even uncurl them while they were still holding a stack of pap-ARRRRGH I’LL DO WHATEVER YOU WANT JUST DON’T HURT THEM PLEASE
That depends entirely on what the PM’s code is. If it doesn’t include input sanitizers, a buffer overflow attack could suffice as a basilisk. If your model of a PM basilisk is “Something that would constitute a logical argument that would harm a PM”, then you’re operating on a very limited understanding of basilisks.