You can never know for certain if your cryptosystem is still safe. But there are few classical ways of going about this.
If you are the one whose cryptosystem may be broken:
Use your cryptosystem to send false information and see if the enemy responds to the false information. If they do, they've read your message.
Note that that indicates that your cryptosystem was broken, but the false information may also have been obtained and leaked in another way - for example, an enemy agent within your own organisation.
If you are the one who broke a cryptosystem:
Every time you use the knowledge obtained by decrypting enemy messages, you risk the enemy becoming aware of yours having broken their crypto.
You need to somehow convince your enemy that you obtained the information in another way. Clearly this is not a trivial task.