Max Nadeau Compares AI Safety Monitoring Arguments

Max Nadeau replied on the topic of AI safety. He described two arguments as similarly structured: one on CoT monitorability and one on the secrecy of frontier algorithms. Nadeau noted that both will probably be lost eventually. He added that the longer they can be maintained the better, since this provides more time to prepare more robust longer-term defenses. All else equal, he favors extending those periods.