Too long to comment, but can you actually replicate that issue?
Because that would appear more of a problem for the sender and not necessarily an incorrect configuration of your own mail servers.
SMTP standards allow and expect you to send an identical message to multiple recipients by issuing multiple RCPT commands before submitting the message and expecting the SMTP server to turn that into unique messages for each of the recipients.
(That's how you efficiently send mails with multiple recipients on the To: CC: and Bcc: lines.)
On my test system (I only have a Postfix and not an Exchange server) including an incorrect address between some valid ones does cause an error on that incorrect address, as expected, but when you continue the messages will still be delivered to all the other valid mailboxes.
I expect Exchange to actually do the same, but can be mistaken.
You can test that quite simply by simulating the message delivery with a telnet session :
telnet mailhost 25
Connected to mailhost
Escape character is '^]'.
220 mailhost ESMTP Postfix (Ubuntu)
helo my.hostname
250 mailhost
mail from: serverfault@my.hostname
250 2.1.0 Ok
rcpt to: valid-mail-address-1
250 2.1.5 Ok
rcpt to: non-existant-mail-address
550 5.1.1 <non-existant-mail-address>: Recipient address rejected: User unknown in local recipient table
At this stage an e-mail client will usually stop sending the message and provide a pop-up to the user requesting them to check the rejected e-mail address, but for debugging purposes you can and should be able to continue with a third, valid, recipient address and then submit an actual email message:
rcpt to: valid-mail-address-2
250 2.1.5 Ok
data
354 End data with <CR><LF>.<CR><LF>
From: ServerFault <serverfault@my.hostname>
Subject: test
This is a test.
.
250 2.0.0 Ok: queued as 7210CBD1AA
quit
221 2.0.0 Bye
That results in a new e-mail message in both of the valid mailboxes, despite the error on the middle recipient, but again I don't have access to an exchange server at the moment.