Re:OpenAI o1 system card
by QuadmasterXLII on Thursday December 05, 2024 @01:34PM
(#42331016)
Keep in mind that if putting “Make sure you achieve your goal in the long term” in the prompt makes the model .3% more effective at whatever task you are giving it, people will absolutely put it in the prompt.
EDIT: Actually the "Oversight subversion" number is 0.3%, I misread the chart. Weight exfiltration is actually 0%, though.