Zephaniah Roe comments on Rerunning AI safety papers on every frontier release would be pretty easy and valuable

[Zephaniah Roe]

This is helpful! I’ll look into the linked example and do some deeper research to try to find analogous work we could be missing.