We should have independent organizations that can access the full details of these training runs for monitoring, not just after various accidents occur.
It's good that OpenAI is sharing this, but the best way for safety is more trust and more eyes on the hard problems to solve.
We have paused some frontier RL training to ensure that we can meet the appropriate alignment, security and monitoring standards for the new level of capabiliti...