There's been a bunch of discourse in the run up to the November AI Safety Summit in the UK about what safety policies should be in place. ARC Evals & Anthropic are pushing for 'Responsible Scaling', which doesn't put any hard upper …