Phantom Update on Aurora Postgres from AWS DMS Replication from SQL Server Source

We have an EC2 instance with SQL Server installed and AWS DMS Replication set up to sync changes from Database A - Table A to an Aurora Postgres Database A Table A. This has been working fine for the last year for inserts/updates/deletes. We recently added a new field (LastUpdateDate) into Table A in SQL Server of type datetimeoffset, this was automatically added to the corresponding Table A in Aurora Postgres of type varchar(34). This works fine for insert and updates. However if the column LastUpdateDate is NULL and we delete this row on the source (SQL Server), an error is throw in DMS where it is attempting to do an update of the corresponding row in Aurora Postgres but trying to set the LastUpdateDate to an invalid date (10000-01-01 04:17:59.0000000 +04:18), this is 35 characters long and so fails to insert into the LastUpdateField which is on 34 characters. My question is we are only doing a delete, where is this update coming from? Where is this strange date coming from? It is not occurring on the SQL Server side as we have no triggers on the table, have cdc enabled and can see that only a Delete operation was done. The delete operation does succeed but we still have these update errors showing up in our 'public.awsdms_apply_exceptions table.'

Any help would be appreciated. Thanks