How does add_loss() behave with composite models?

Solution for How does add_loss() behave with composite models?
is Given Below:

Suppose I have a base model:

model1 = keras.Model(inp,outp)

And I use add loss:

model1.add_loss( myCustomLoss(...))

Then I make a second model that incorporates model1, something like this:

model_composite = keras.Model(inp2, model2(model1(inp2))
model_composite.compile(..., loss=anotherLoss)

Does the add_loss (myCustomLoss) come along for the ride in model_composite? If so, is there a way for that to be disabled so I can train model1 alone with the add_loss, but not have it be part of the composite training?