Mitigations suggested for AI control often overlap with those that responsible AI developers already implement to ensure robustness.