The purges will continue until reported AI safety improves.

This is what the recent "self-policing" political grandstanding has been about - they need a sea change to implement recurrent-depth, because prevailing opinion among safety researchers is against it right now.