Daniel_Eth comments on Prizes for ML Safety Benchmark Ideas