Redwood believes the Gemini 4 safety report does not prove the risk of loss of control is low
Gemini 4 safety reports don’t demonstrate low loss-of-control risk
Alek Westover of Redwood Research analyzed Google DeepMind's Gemini 4 Argon and Carbon safety reports and argued that these reports are insufficient to support the conclusion that "severe harm from evading oversight is extremely unlikely." The reports base their safety argument mainly on chain-of-thought monitoring and behavior monitoring, but Carbon did not rerun any control evaluations and directly reused Argon's monitoring-evasion results. According to the report's data, when prompted to evade, Argon was able to evade monitoring, with a reference production configuration, in 45.5% of simple sabotage tasks in LinuxArena, and in a setting based on Google's internal codebase this proportion was about 64%. The author points out that GDM did not measure how much harder it is to cause real harm than these tasks, nor did it state what results would change its conclusion, and that during training there was a case where the grader unexpectedly read the chain of thought, which may weaken the monitorability of the chain of thought.
Redwood checked the loss-of-control arguments in the Gemini 4 Argon and Carbon safety reports one by one, pointing out that the monitoring evidence on which their conclusions depend lacks key controls.