open question - handling of transactionId in case of reorg.

I just browsed through the code in 5 minutes and i have a bad feeling:

confirmTransaction only takes a transactionId - and not a hash of the underlying semantics. / tx summary.

what if there is a 2-of-2 ms wallet , user A, B. A submits a harmless tx, asks B to confirm. B confirms transactionId 1. Meanwhile, user A (who might own some mining power) publishes a malicious tx also with transactionId 1, and includes it on a different branch. reorg happens now on that branch user B agrees to the malicious tx.

can this be mitigated by user B in any way? - other than waiting a certain # of confirmations?