But if AI companies don’t develop organizational practices that allow them to reliably prevent analogous problems like this before they affect the training and deployment of several frontier models
Hey, sorry, new here so apologies if this is covered elsewhere. What could these organizational practices look like? Sounds like you agree an automated CoT monitoring system is a reasonable defense, but does this look like:
(1) more partnering with independent auditors pre-release / external redteaming
(2) just more safety gates and stronger enforcement
Hey, sorry, new here so apologies if this is covered elsewhere. What could these organizational practices look like? Sounds like you agree an automated CoT monitoring system is a reasonable defense, but does this look like:
(1) more partnering with independent auditors pre-release / external redteaming
(2) just more safety gates and stronger enforcement
or something different?