Atomic insert operations in SPARQL

So, I have the following requirement: create a new instance of type A with label X, where X is unique for all instances of A.

Initially the way I implemented this feature was to execute an ASK query to check if an instance of A with rdfs:label X already exists. If so, ask the user for a unique label, otherwise execute the INSERT query. The problem with this approach is that the ASK query and the INSERT query are not atomic operations (I am currently using TDB/Fuseki for the triple store). During the time the ASK query is executed and the time the INSERT query is executed, a new instance of A could be created with the same label X.

What would be the best approach to make sure the uniqueness check and insert are an atomic operation?

What I have thought about is to do the uniqueness check and creation in the same query by having a filter in the Where clause, that would execute the insert only if the label is unique. After this query is executed, then I would execute another ASK query to check if the instance was created (if the instance was not created, then I know there is a duplicate label). The query would look something like this:

INSERT 
{
  ?newInstance a xyz:A .
  ?newInstance rdfs:label 'label here'^^xsd:string .
}
WHERE 
{   
    ?existingInstance a xyz:A  .
    ?existingInstance rdfs:label ?existingInstanceLabel .
    BIND (IRI(<xyz.com/instances/newInstance>) AS ?newInstance ) .
    FILTER (!regex(?existingInstanceLabel ,"^label here$","i")) .
}

This query would not work since the WHERE clause would get executed for any instance that does not match the new label.