When the safety community designs their systems to also maintain security properties, it is likely that public-key encryption will be among the tools that are applied. The security guarantees of this technology are based on a particular model of computation. We present the properties of this model that are relevant in the setting of distributed systems. Of particular importance is that the model has no notion of time. From this it follows that systems that need to be available must exercise the utmost care before applying public-key encryption in any form. We discuss the relation between public-key encryption and timeliness, the tradeoffs that must be made at design time, and how the property of (lack of) availability might very well contaminate other system components.