It's not really clear to me what harm could actually be caused by doing this, if it's possible. Could you describe a more detailed scenario of concern in the question?

Let's say I send you some private message, let's say a bug report about a vulnerability.

Now, Mallory wants to read that message, and also your reply to it. He may take that message, add himself as Cc (AFAIK, the Cc field is not a protected field; see crypt_protected_headers_read in muttrc(5)), and modify the metadata of the encrypted message to add his keyID as a recipient, even when he's not able to decrypt the message.

When you receive my bug report, altered by Mallory, you'll be lead to think that I CCed him, and that I want him to be part of the conversation, and so in your reply, which will likely contain (at least part of) my original report, you will likely encrypt it for both me and Mallory and send it to both.

So, is Mallory able to modify the metadata of an encrypted message to add false recipients to it?

The critical point of your question is, how would Mallory do this: " He may take that message, add himself as Cc". This needs to be done during encryption - if Mallory is in a position to do that, you already have a security breach. If Mallory intercepted the already encrypted message, he could do nothing useful with it.

aheinlein‭

" He may take that message, add himself as Cc": By "message" here I meant a RFC822 message, that is, an email. The header of an email is sent in plain text, and a MITM can easily modify it.